WO2018054330A1 - 数据处理方法、装置和存储介质 - Google Patents

数据处理方法、装置和存储介质 Download PDF

Info

Publication number
WO2018054330A1
WO2018054330A1 PCT/CN2017/102702 CN2017102702W WO2018054330A1 WO 2018054330 A1 WO2018054330 A1 WO 2018054330A1 CN 2017102702 W CN2017102702 W CN 2017102702W WO 2018054330 A1 WO2018054330 A1 WO 2018054330A1
Authority
WO
WIPO (PCT)
Prior art keywords
data
layer
processing
sample
data combination
Prior art date
Application number
PCT/CN2017/102702
Other languages
English (en)
French (fr)
Inventor
李英杰
万乐
邓大付
殷俊
程序
杜家春
Original Assignee
腾讯科技（深圳）有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 腾讯科技（深圳）有限公司 filed Critical 腾讯科技（深圳）有限公司
Priority to EP17852398.1A priority Critical patent/EP3518105B1/en
Publication of WO2018054330A1 publication Critical patent/WO2018054330A1/zh
Priority to US16/221,842 priority patent/US11135514B2/en

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A63—SPORTS; GAMES; AMUSEMENTS
    • A63F—CARD, BOARD, OR ROULETTE GAMES; INDOOR GAMES USING SMALL MOVING PLAYING BODIES; VIDEO GAMES; GAMES NOT OTHERWISE PROVIDED FOR
    • A63F13/00—Video games, i.e. games using an electronically generated display having two or more dimensions
    • A63F13/55—Controlling game characters or game objects based on the game progress
    • A63F13/57—Simulating properties, behaviour or motion of objects in the game world, e.g. computing tyre load in a car race game
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/54—Interprogram communication
    • G06F9/542—Event management; Broadcasting; Multicasting; Notifications
    • A—HUMAN NECESSITIES
    • A63—SPORTS; GAMES; AMUSEMENTS
    • A63F—CARD, BOARD, OR ROULETTE GAMES; INDOOR GAMES USING SMALL MOVING PLAYING BODIES; VIDEO GAMES; GAMES NOT OTHERWISE PROVIDED FOR
    • A63F13/00—Video games, i.e. games using an electronically generated display having two or more dimensions
    • A63F13/30—Interconnection arrangements between game servers and game devices; Interconnection arrangements between game devices; Interconnection arrangements between game servers
    • A63F13/35—Details of game servers
    • A—HUMAN NECESSITIES
    • A63—SPORTS; GAMES; AMUSEMENTS
    • A63F—CARD, BOARD, OR ROULETTE GAMES; INDOOR GAMES USING SMALL MOVING PLAYING BODIES; VIDEO GAMES; GAMES NOT OTHERWISE PROVIDED FOR
    • A63F13/00—Video games, i.e. games using an electronically generated display having two or more dimensions
    • A63F13/40—Processing input control signals of video game devices, e.g. signals generated by the player or derived from the environment
    • A63F13/42—Processing input control signals of video game devices, e.g. signals generated by the player or derived from the environment by mapping the input signals into game commands, e.g. mapping the displacement of a stylus on a touch screen to the steering angle of a virtual vehicle
    • A—HUMAN NECESSITIES
    • A63—SPORTS; GAMES; AMUSEMENTS
    • A63F—CARD, BOARD, OR ROULETTE GAMES; INDOOR GAMES USING SMALL MOVING PLAYING BODIES; VIDEO GAMES; GAMES NOT OTHERWISE PROVIDED FOR
    • A63F13/00—Video games, i.e. games using an electronically generated display having two or more dimensions
    • A63F13/40—Processing input control signals of video game devices, e.g. signals generated by the player or derived from the environment
    • A63F13/44—Processing input control signals of video game devices, e.g. signals generated by the player or derived from the environment involving timing of operations, e.g. performing an action within a time slot
    • A—HUMAN NECESSITIES
    • A63—SPORTS; GAMES; AMUSEMENTS
    • A63F—CARD, BOARD, OR ROULETTE GAMES; INDOOR GAMES USING SMALL MOVING PLAYING BODIES; VIDEO GAMES; GAMES NOT OTHERWISE PROVIDED FOR
    • A63F13/00—Video games, i.e. games using an electronically generated display having two or more dimensions
    • A63F13/45—Controlling the progress of the video game
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/54—Interprogram communication
    • G06F9/545—Interprogram communication where tasks reside in different layers, e.g. user- and kernel-space

Definitions

  • Embodiments of the present invention relate to the field of data processing, and in particular, to a data processing method, apparatus, and storage medium.
  • the data processing rules for turn-based events are relatively simple.
  • the objects used for event execution have global event information for the events, and take turns to act, have a longer decision time for the events, and the event behaviors are immediately implemented, and the feedback is determined.
  • the turn-based event is a turn-based game.
  • the rules are relatively simple.
  • the game player has global information, takes turns to act, has a long decision time, and the behavior can be implemented immediately, or can be determined through feedback, for example, Go.
  • the data processing rules for real-time events are complex.
  • the object used for executing the event has only part of the event information, and the action is performed.
  • the decision time for the event is short, the event behavior needs time execution and has a certain probability of success. Therefore, the real-time event There is a difference between data processing and data processing for turn-based events.
  • the AlphaGo is an algorithm that uses deep learning training strategy networks and value networks, and uses Monte Carlo tree integration to achieve high-level Go moves.
  • 1 is a schematic structural diagram of a decision network and a value network according to an AlphaGo algorithm in the related art.
  • the AlphaGo algorithm uses a deep learning training strategy network and a value network.
  • Human expert positions pass the Strategic Network (SL Network) on the human expert side through the Classification Outward Policy. Policy Gradient is transmitted to the Self-play Positions.
  • SL Network Strategic Network
  • Policy Gradient is transmitted to the Self-play Positions.
  • training is performed through the RL Network and the Value Network on the artificial intelligence side to obtain data.
  • the policy network on the human expert side and the policy network on the artificial intelligence side are collectively referred to as data.
  • the Policy Network, the Policy Network and the Value Network are trained by certain algorithm formulas and implemented by Monte Carlo Tree Search (MCTS) algorithm.
  • MCTS Monte Carlo Tree Search
  • FIG. 2 is a schematic diagram of a Monte Carlo tree search algorithm according to the related art. As shown in Fig. 2, the probability of selecting the drop probability, expanding the sub-samples through the policy network, evaluating the current pass income through the value network, and feeding back the current travel revenue, and selecting the strategy network through the Monte Carlo search algorithm Probability and value network assessment of the current proceeds of the integration simulation, and finally select the best placement based on the current disk.
  • the data processing complexity of real-time events is much higher than the data processing complexity of the above-mentioned turn-based events. Because there are many differences between the data processing of the turn-based events and the data processing of real-time events, the combination of the two-layer network of the algorithm cannot be satisfied. Real-time event macro-decision needs are met, and it is impossible to meet the micro-operation level, which cannot meet the needs of real-time event intelligent systems, and the data processing efficiency is low.
  • the embodiment of the invention provides a data processing method, device and storage medium to solve at least the technical problem of low data processing efficiency of the related art.
  • a data processing method includes: obtaining sample data of a game client execution event; performing pre-processing on the sample data to obtain a multi-layer data combination, wherein each layer of the data combination in the multi-layer data combination corresponds to one target of the same target event
  • the event object, the different layer data combination in the multi-layer data combination corresponds to different target event objects in the same target event; the processing is performed on each layer of data combination according to the preset processing algorithm, and the processing result of each layer data combination is obtained;
  • Combined processing results obtains a target instruction, wherein the target instruction is used to instruct the game client to simultaneously execute different target event objects corresponding to different layer data combinations.
  • the data processing apparatus includes: a first obtaining unit configured to acquire sample data of a game client execution event; and a first processing unit configured to perform pre-processing on the sample data to obtain a multi-layer data combination, wherein the multi-layer data
  • a first obtaining unit configured to acquire sample data of a game client execution event
  • a first processing unit configured to perform pre-processing on the sample data to obtain a multi-layer data combination, wherein the multi-layer data
  • Each layer of data combination in the combination corresponds to one target event object in the same target event, and the different layer data combinations in the multi-layer data combination correspond to different target event objects in the same target event, and the target event object is to be held on the game client.
  • the event object executed at the same time; the second processing unit is configured to perform processing on each layer of data combination according to a preset processing algorithm to obtain a processing result of each layer of data combination; and the third processing unit is set to combine data for each layer
  • the processing result is integrated to obtain a target instruction, wherein the target instruction is used to instruct the game client to simultaneously execute different target event objects corresponding to different layer data combinations.
  • a terminal wherein the terminal is arranged to execute program code for performing the steps in the data processing method of the embodiment of the present invention.
  • a storage medium configured to store program code for performing the steps in the data processing method of the embodiment of the present invention.
  • the sample data of the execution event of the game client is acquired; the pre-processing is performed on the sample data to obtain a multi-layer data combination, and each layer of the data combination in the multi-layer data combination corresponds to one target in the same target event.
  • the event object, the different layer data combination in the multi-layer data combination corresponds to different target event objects in the same target event, and the target event object is an event object to be executed simultaneously on the game client; the data combination of each layer is performed according to a preset processing algorithm.
  • the purpose of the instruction is to achieve the technical effect of improving the efficiency of data processing, thereby solving the technical problem of low data processing efficiency of the related art.
  • FIG. 1 is a schematic structural diagram of a decision network and a value network according to an AlphaGo algorithm in the related art
  • FIG. 2 is a schematic diagram of a Monte Carlo tree search algorithm according to the related art
  • FIG. 3 is a schematic diagram of a hardware environment of a data processing method according to an embodiment of the present invention.
  • FIG. 4 is a flow chart of a data processing method according to an embodiment of the present invention.
  • FIG. 5 is a flowchart of a method of marking sample data according to a plurality of sample sequences of sample data, in accordance with an embodiment of the present invention
  • FIG. 6 is a flow chart of a method of performing pre-processing on tagged sample data in accordance with an embodiment of the present invention
  • FIG. 7 is a flow chart of another data processing method according to an embodiment of the present invention.
  • FIG. 8 is a flowchart of another data processing method according to an embodiment of the present invention.
  • FIG. 9 is a flowchart of a method of performing processing on sample information in each layer of data combination according to a processing algorithm corresponding to each layer of data combination in a combination of multiple layers of data according to an embodiment of the present invention
  • FIG. 10 is a flowchart of another data processing method according to an embodiment of the present invention.
  • FIG. 11 is a flowchart of another data processing method according to an embodiment of the present invention.
  • FIG. 12 is a schematic diagram of an interaction process in a game process according to an embodiment of the present invention.
  • FIG. 13 is a flow chart of another method of marking sample data based on a plurality of sample sequences of sample data, in accordance with an embodiment of the present invention.
  • FIG. 14 is a flowchart of another data processing method according to an embodiment of the present invention.
  • 15 is a flow chart showing a method of game interaction according to an embodiment of the present invention.
  • 16 is a schematic diagram of a data processing apparatus according to an embodiment of the present invention.
  • FIG. 17 is a schematic diagram of another data processing apparatus according to an embodiment of the present invention.
  • FIG. 18 is a schematic diagram of another data processing apparatus according to an embodiment of the present invention.
  • FIG. 19 is a schematic diagram of another data processing apparatus according to an embodiment of the present invention.
  • FIG. 20 is a schematic diagram of another data processing apparatus according to an embodiment of the present invention.
  • 21 is a schematic diagram of another data processing apparatus according to an embodiment of the present invention.
  • FIG. 22 is a schematic diagram of another data processing apparatus according to an embodiment of the present invention.
  • FIG. 23 is a schematic diagram of another data processing apparatus according to an embodiment of the present invention.
  • FIG. 24 is a structural block diagram of a terminal according to an embodiment of the present invention.
  • an embodiment of a data processing method is provided.
  • the data processing method described above may be applied to a hardware environment composed of the server 302 and the terminal 304 as shown in FIG. 3.
  • 3 is a schematic diagram of a hardware environment of a data processing method according to an embodiment of the present invention.
  • the server 302 is connected to the terminal 304 through a network.
  • the network includes but is not limited to a wide area network, a metropolitan area network, or a local area network.
  • the terminal 304 is not limited to a PC, a mobile phone, a tablet, or the like.
  • the data processing method of the embodiment of the present invention may be executed by the server 302, may be performed by the terminal 304, or may be jointly performed by the server 302 and the terminal 304.
  • the data processing method performed by the terminal 304 in the embodiment of the present invention may also be performed by a client installed thereon.
  • the data processing method may include the following steps:
  • Step S402 acquiring sample data of a game client execution event.
  • step S402 of the present application sample data of a game client execution event is acquired.
  • the game client is used to execute an event, for example, to execute an event in a human-machine battle mode in a Real Time Game, which is different from a turn-based game.
  • the game client of this embodiment generates data during the execution of the event, and the data may be game data, and the sample data is data of a part of the actual observation or investigation during the execution of the event by the game client, which may be randomly selected. , with enough data to reflect the overall situation of the game client's execution of the event.
  • This embodiment can also reasonably construct the data of the game client in the process of executing the event through the multi-layer deep learning framework, thereby obtaining sample data.
  • sample data of a game client execution event where the sample data is an input sample, which may be a game sample, the game sample includes a plurality of sample sequences, and the plurality of sample sequences have different priorities, wherein different priorities
  • the sample sequence of the level may include the same data frame.
  • Step S404 performing preprocessing on the sample data to obtain a multi-layer data combination.
  • preprocessing is performed on the sample data to obtain a multi-layer data combination, wherein each layer of the data combination in the multi-layer data combination corresponds to one target event object in the same target event, and more The different layer data combinations in the layer data combination correspond to different target event objects in the same target event, and the target event object is an event object to be executed simultaneously on the game client.
  • the design data of the game data is numerous and cannot be directly used as training data
  • a plurality of sample sequences of the sample data are acquired, and the sample data is marked according to the plurality of sample sequences to obtain the labeled sample.
  • the sample data can be marked by preset tag logic.
  • the entire sample is marked according to a preset logic configuration, thereby obtaining labeled sample data.
  • the priority of multiple sample sequences may be separated according to the characteristics of the event.
  • the different sample sequences are marked according to the priority order of the sample sequence, and the sample sequence can be marked by using a preset rule or a preset sample segmentation algorithm to obtain a mark frame.
  • the traversal of all the marked frames, the adjacent identical marked frames are marked as a sample sequence, and the start frame and the end frame of each sample sequence are marked, thereby obtaining the labeled sample sequence data.
  • the multi-layer data combination of this embodiment can be assembled from the state information extracted by the general state function in the labeled sample data.
  • the state information of the current disk surface can be extracted from the mark sample data using a general state function, and the state is Information is assembled into a multi-layer data combination.
  • the target event object is an event object to be executed simultaneously on the game client, and can correspond to the character status, the friendly character status, the enemy damage information, the map information, and the non-player control character (Non Player Controlled Character, NPC for short). ) Information, etc.
  • Each layer of data combination in the multi-layer data combination of this embodiment corresponds to one target event object in the same target event, and different layer data combinations in the multi-layer data combination correspond to the same target event.
  • the different target event objects that is, the data combination of each layer in the multi-layer data combination has a one-to-one correspondence with the target event objects in the same target event, for example, the target event is a competition event, and the target event object includes to be executed simultaneously.
  • the B event object in the event, the third layer data combination corresponds to the C event object in the battle event.
  • the event objects to be executed simultaneously include the local character, the friendly character, the enemy, and the non-player control role, and each layer of data combination corresponds to the local character, the friendly character, the enemy, the non-player control role, and the like.
  • the different layer data combinations in the multi-layer data combination of this embodiment correspond to different target event objects in the same target event. For example, in a real-time game of five parties, the role status of the party is used as the first layer, and the friend is the friend.
  • the party role status is ranked as the second to fifth layers according to the strength ranking, that is, the second to fifth layers distinguish the friendly characters due to the different rankings of the strengths, and the enemy ranks the sixth to the tenth according to the damage ranking.
  • Layers that is, the sixth to tenth layers distinguish the enemy due to the difference in damage ranking, map information and NPC information as the outermost layer, and other data assembly principles are also applicable, and there is no limitation here.
  • the state information of the sample data in each layer of data combination, each character data is mapped to a certain legal action space according to the game rule state, thereby obtaining event data.
  • the status information, each role data, and each event data in each layer of data combination are composed of sample information in each layer of data combination.
  • Performing rotation processing on the sample data to expand the sample number corresponding to the sample data, and other user information may also be added to the sample information, for example, adding the error rate of the execution event and the operation frequency to the sample information, thereby facilitating training and realizing the label sample.
  • the data is pre-processed to obtain the purpose of multi-layer data combination.
  • Step S406 performing processing on each layer of data combination according to a preset processing algorithm to obtain a processing result of each layer of data combination.
  • step S406 of the present application processing is performed on each layer of data combination according to a preset processing algorithm, and a processing result of each layer of data combination is obtained.
  • Each layer of data combination in the multi-layer data combination has a corresponding preset processing algorithm, and the preset processing algorithm may be to learn the current state information according to the sample information in each layer of data combination.
  • the probability model of the execution probability and the value model of the execution value of the event for example, learning the current state according to the state information of the current disk surface in the sample, the role information, and the role mapping according to the game rule state to a corresponding value in a legal action space.
  • the lower action performs the probability model and the value model of the action.
  • the specific algorithm can be the AlphaGo integrated strategy network and the Monte Carlo tree search algorithm of the value network. Among them, the Monte Carlo tree search algorithm is a heuristic search algorithm for decision-making.
  • the processing result of each layer of data combination is obtained.
  • the result is used for the decision layer output to perform integration processing, which is a result obtained by performing processing on each layer of data combination according to a preset processing algorithm, and the processing result includes an execution probability and an execution value, and may be preset according to a preset probability corresponding to each layer of data combination.
  • the model and the preset value model perform processing on the sample information in each layer of data combination, and obtain the execution probability and execution value of the target event of the game client corresponding to each layer of data combination, for example, the preset probability model is an action execution probability model.
  • the preset value model is a value model of the action, and the sample information in each layer of the data combination is processed according to the action execution probability model and the action value model in the current state, and the game client execution target event corresponding to each layer of the data combination is obtained. Execution probability and execution value.
  • Step S408 performing integration processing on the processing result of each layer of data combination to obtain a target instruction.
  • the processing result of each layer of data combination is integrated to obtain a target instruction, wherein the target instruction is used to instruct the game client to simultaneously execute different target events corresponding to different layer data combinations.
  • the processing result of each layer of data combination is weighted and integrated to obtain a target instruction, and the target instruction is also the final strategy. Instructing the game client to simultaneously execute different target event objects corresponding to different layer data combinations.
  • the target instruction can be added to the state evaluation function to determine whether the target instruction needs to be changed under the current disk surface to satisfy the variable event execution environment.
  • the target instruction is executed, that is, the final strategy is executed.
  • the current disk surface information of the game client may be displayed, and the current disk surface information is an execution result of the instruction to the target instruction, and is determined according to the preset state evaluation function and the current execution result.
  • the behavior tree is a graphical model language, which is used in games to describe the execution conditions and methods of different behaviors, so as to ensure the fast execution of behaviors to enhance the experience of game players.
  • the sample data of the execution event of the game client is acquired; the preprocessing is performed on the sample data to obtain a multi-layer data combination, and each layer of the data combination in the multi-layer data combination corresponds to one of the same target events.
  • the target event object, the different layer data combination in the multi-layer data combination corresponds to different target event objects in the same target event, and the target event object is an event object to be executed simultaneously on the game client; according to each combination with the multi-layer data
  • the processing algorithm corresponding to the layer data combination performs processing on the sample information in each layer of data combination to obtain a processing result of each layer of data combination; and integrates the processing result of each layer of data combination to obtain a target instruction, wherein the target instruction is used for Instructing the game client to simultaneously execute different target event objects corresponding to different layer data combinations can solve the technical problem of low data processing efficiency of the related technology, thereby achieving the technical effect of improving data processing efficiency.
  • step S404 performing pre-processing on the sample data, and obtaining the multi-layer data combination includes: marking the sample data according to the plurality of sample sequences of the sample data to obtain the labeled sample data; and performing the marking sample data
  • Pre-processing results in a multi-layer data combination in which different layer data combinations in a multi-layer data combination correspond to different processing algorithms and different sample information.
  • the multi-layer data combination uses different algorithms, which can make a compromise between the decision time and the simulation depth to meet the varied game scenarios, and can cope with different decision-making time requirements, so that the decision-making execution is simple and efficient, thus ensuring the behavior. Fast execution.
  • the processing algorithm corresponding to each layer of data combination in the multi-layer data combination may be used.
  • the processing of the sample information in each layer of data combination is performed to obtain the processing result of each layer of data combination.
  • step S404 the sample data is marked according to the plurality of sample sequences of the sample data, and obtaining the labeled sample data includes: sequentially passing the marking according to the priority of each of the plurality of sample sequences.
  • the frame marks each sample sequence to obtain a plurality of labeled sample sequences, and then combines the adjacent labeled sample sequences of the plurality of labeled sample sequences according to the same marked frame to obtain a starting frame and a ending in the combined labeled sample sequence.
  • the frame is marked to obtain the labeled sample data.
  • FIG. 5 is a flow chart of a method of tagging sample data based on a plurality of sample sequences of sample data, in accordance with an embodiment of the present invention.
  • the method for marking sample data according to a plurality of sample sequences of sample data includes the following steps:
  • Step S501 determining a priority of each of the plurality of sample sequences.
  • the priority of each of the plurality of sample sequences is determined.
  • the sample data includes a plurality of sample sequences, and each of the plurality of sample sequences has a priority. Since the event execution rule is complicated, a case where the same data frame belongs to multiple sample sequences occurs, and sample data of the execution event of the game client is acquired. Thereafter, the priority of each of the plurality of sample sequences is determined, and the order of the plurality of sample sequences is derived according to the priority of each sample sequence.
  • Step S502 marking each sample sequence by marker frames in order of priority, to obtain a plurality of marker sample sequences.
  • each sample sequence is marked by a mark frame in order of priority to obtain a plurality of mark sample sequences.
  • the different sample sequences can be marked by the mark frame in the order in which the plurality of sample sequences are arranged.
  • a plurality of labeled sample sequences are obtained by marking a frame by marking a different sequence of samples according to an arrangement order of the plurality of sample sequences by using a preset rule or a preset sample segmentation algorithm.
  • Step S503 the adjacent tag sample sequences in the plurality of tag sample sequences are the same
  • the marker frames are merged to obtain a merged marker sample sequence.
  • step S503 of the present application after each sample sequence is sequentially marked by a mark frame according to priority, a plurality of mark sample sequences are obtained, and adjacent mark sample sequences in the plurality of mark sample sequences are followed.
  • the same marker frames are merged to obtain a merged marker sample sequence.
  • the tagged frames of the plurality of labeled sample sequences may be traversed, and the sample sequences of adjacent identical labeled frames may be marked as the same sequence to obtain a merged labeled sample sequence.
  • Step S504 marking the start frame and the end frame of the merged mark sample sequence to obtain mark sample data.
  • the start frame of the merged tag sample sequence is obtained.
  • the end frame is marked to obtain the labeled sample data.
  • the labeling sample data may be obtained by combining the adjacent labeling sample sequences in the plurality of label sample sequences according to the same labeling frame to obtain the combined labeling sample sequence, and marking the starting frame and the ending frame of each of the combined labeling sample sequences.
  • the embodiment determines the priority of each sample sequence in the plurality of sample sequences; sequentially marks each sample sequence by the mark frame in order of priority to obtain a plurality of mark sample sequences; and adjacent ones of the plurality of mark sample sequences
  • the labeled sample sequence is merged according to the same marked frame to obtain a merged labeled sample sequence; the start frame and the end frame of the merged labeled sample sequence are marked to obtain marked sample data, and sample data of multiple sample sequences according to the sample data is realized. Marking is performed for the purpose of marking the sample data.
  • step S404 performing pre-processing on the tag sample data, and obtaining the multi-layer data combination comprises: performing assembly on the game client executing different state information of the current event object, and obtaining a multi-layer data combination.
  • FIG. 6 is a flow chart of a method of performing pre-processing on tagged sample data in accordance with an embodiment of the present invention. As shown in FIG. 6, the method for performing preprocessing on the labeled sample data includes the following steps:
  • Step S601 extracting a game client from the mark sample data by using a preset state function Lines different state information for the current event object.
  • step S601 of the present application after the sample data is marked according to the plurality of sample sequences of the sample data, and the labeled sample data is obtained, different states of the current disk surface are extracted in the labeled sample data by using a general state function.
  • the different status information may be used to indicate status information of the role of the game client in the real-time game process, the role of the friendly party, the state of the enemy role, and the like, and the current event object is the event object currently executed by the game client. .
  • Step S602 performing assembly on different state information to obtain a multi-layer data combination.
  • the different state information is assembled, for example, The state information such as the party role state, the friendly role state, and the enemy character state are assembled, and a multi-layer data combination is obtained.
  • Each layer of the data combination in the multi-layer data combination corresponds to one target event object in the same target event, and multiple layers.
  • the different layer data combinations in the data combination correspond to different target event objects in the same target event, and the target event objects are event objects to be executed simultaneously on the game client.
  • the role status of the party is the first level
  • the status of the friendly role is placed on the second to fifth levels according to the strength ranking
  • the enemy role status is ranked according to the damage ranking.
  • the outermost layer is map information and NPC information, and other data assembly methods are also applicable.
  • the embodiment extracts different state information of the current event object by the game client in the mark sample data by using a preset state function, and the current event object is an event object currently executed by the game client; performing assembly on different state information, and obtaining more
  • the layer data combination realizes the preprocessing of the labeled sample data, and the purpose of obtaining the multi-layer data combination, thereby improving the data processing efficiency.
  • processing is performed on the sample information in each layer of data combination according to a processing algorithm corresponding to each layer of data combination in the combination of the multi-layer data to obtain each layer of data groups.
  • sample information is generated from the status information, the character data on the game client, and the event data of the target event.
  • FIG. 7 is a flow chart of another data processing method in accordance with an embodiment of the present invention. As shown in FIG. 7, the data processing method includes the following steps:
  • Step S701 acquiring character data on the game client.
  • the game client has role data when the target event is executed, and the role data is used to represent data of the virtual application body that executes the target event.
  • the character data includes a plurality of character data, and each character data corresponds to data of one virtual application body, and each character data on the game client is acquired.
  • step S702 the state information and the role data are mapped to the preset processing model according to the preset mapping system, and event data of the target event is obtained.
  • the state information and the role data may be mapped into a legal action space through the game rule state, and the action space has event data of the target event corresponding to the state information and the role data,
  • the event data may be action data, thereby obtaining event data of the target event through the status information and the role data and the legal action space.
  • Step S703 generating sample information according to the status information, the role data, and the event data.
  • the sample is generated according to the state information, the role data, and the event data.
  • Information the sample information includes frame information, and each event data corresponds to one piece of sample information.
  • status information, role data, and event data are represented by ⁇ S, u, a>, where S is used to represent status information, u is used to represent role data, and a is used to represent event data.
  • the embodiment obtains the role data on the game client; maps the state information and the role data to the preset processing model according to the preset mapping system, and obtains event data of the target event; and generates sample information according to the state information, the role data, and the event data,
  • the processing algorithm corresponding to each layer of data combination performs processing on the sample information in each layer of data combination, and obtains the processing result of each layer of data combination, thereby improving the efficiency of data processing.
  • the preset information is added to the sample information of the sample size.
  • FIG. 8 is a flow chart of another data processing method in accordance with an embodiment of the present invention. As shown in FIG. 8, the data processing method further includes the following steps:
  • Step S801 performing rotation processing on the sample data to expand the number of samples corresponding to the sample data.
  • step S801 of the present application after the sample information is generated based on the state information, the character data, and the event data, a rotation process is performed on the sample data to expand the number of samples corresponding to the sample data.
  • Step S802 adding preset information to the sample information of the sample size.
  • the preset information is added to the sample information of the sample size.
  • the preset information may be information of other users, such as a failure rate, an operation frequency, and the like, and the information such as the error rate and the operation frequency is added to the frame information, so as to train the personalized strategy.
  • step S406 processing is performed on the sample information in each layer of data combination according to a processing algorithm corresponding to each layer of data combination in the multi-layer data combination, to obtain a processing result of each layer of data combination.
  • the method includes: performing, according to a preset probability model corresponding to each layer of data combination, a preset value model, performing processing on sample information in each layer of data combination, and obtaining an execution probability and execution of a target event of the game client corresponding to each layer of data combination.
  • the value, and the target instruction is obtained by the execution probability corresponding to each layer of data combination and the execution value corresponding to each layer of data combination.
  • FIG. 9 is a flow chart of a method for performing processing on sample information in each layer of data combination according to a processing algorithm corresponding to each layer of data combination in a combination of multiple layers of data according to an embodiment of the present invention.
  • the method includes the following steps:
  • Step S901 Perform processing on the sample information in each layer of data combination according to a preset probability model corresponding to each layer of data combination, and obtain an execution probability of the target client execution target event corresponding to each layer of data combination.
  • each layer of data combination corresponds to a preset probability model, and each layer of data combination learns the preset probability model of the action execution in the current state according to the sample information of each layer of data combination, and obtains each layer The execution probability of the target event performed by the game client corresponding to the data combination.
  • Step S902 performing processing on the sample information in each layer of data combination according to a preset value model corresponding to each layer of data combination, and obtaining an execution value of the target event of the game client corresponding to each layer of data combination.
  • each layer of data combination corresponds to a preset value model, and each layer of data combination learns a preset value model executed by an action in a current state according to sample information of each layer of data combination, and obtains each layer The value probability of the target client executing the target event corresponding to the data combination.
  • Step S903 the execution probability corresponding to each layer of data combination and the execution value corresponding to each layer of data combination are integrated to obtain a target instruction.
  • step S903 of the present application after the game client corresponding to each layer of data combination obtains the execution probability of the target event, and the game client corresponding to each layer of data combination performs the execution value of the target event, The execution probability corresponding to each layer of data combination and the execution value corresponding to each layer of data combination are integrated, the target instruction is obtained, the final strategy is output, and the state evaluation function is added during the strategy, thereby determining whether the current disk needs to be changed. To respond to a changing event execution environment.
  • an execution probability of the target event of the game client corresponding to each layer of data combination is obtained;
  • the default value model corresponding to the data combination for each layer The processing is performed according to the sample information in the combination, and the execution value of the execution target event of the game client corresponding to each layer of data combination is obtained, and the processing algorithm corresponding to each layer of the data combination in the multi-layer data combination is implemented to combine each layer of data.
  • the sample information in the execution process is processed to obtain the processing result of each layer of data combination, and the execution probability corresponding to each layer of data combination and the execution value corresponding to each layer of data combination are integrated to obtain a target instruction, thereby improving data processing efficiency.
  • the target instruction is updated if the target instruction needs to be updated.
  • FIG. 10 is a flow chart of another data processing method in accordance with an embodiment of the present invention. As shown in FIG. 10, the data processing method includes the following steps:
  • Step S1001 Determine, according to the preset state evaluation function, whether the target instruction needs to be updated.
  • the processing result of each layer of data combination is obtained, and the processing result of each layer of data combination is integrated, and after the target instruction is obtained, the target event is executed according to the target instruction, and the corresponding event is returned.
  • the disk information determines whether the target instruction needs to be updated according to the preset state evaluation function, and can determine whether the target instruction needs to be updated through the behavior tree.
  • step S1002 if it is determined that the target instruction needs to be updated, the target instruction is updated.
  • step S1002 of the present application after determining whether the target instruction needs to be updated according to the preset state evaluation function, if it is determined that the target instruction needs to be updated, the target instruction is updated, thereby coping with the variable event processing environment.
  • the target instruction is obtained by integrating the processing result of each layer of data combination, it is determined according to the preset state evaluation function whether the target instruction needs to be updated; if it is determined that the target instruction needs to be updated, the target instruction is updated and improved. Data processing efficiency.
  • the processing of each layer of data combination is updated according to different target state information when the target object is executed with different target event objects.
  • an update of the data combination for each layer is obtained.
  • the processing result is obtained by integrating the update processing result of the multi-layer data combination to obtain an update target instruction.
  • FIG. 11 is a flow chart of another data processing method in accordance with an embodiment of the present invention. As shown in FIG. 11, the data processing method includes the following steps:
  • Step S1101 Acquire different target state information when the game client executes different target event objects according to the target instruction.
  • the processing result of each layer of data combination is obtained, and the processing result of each layer of data combination is integrated, and after the target instruction is obtained, the game client executes different targets according to the target instruction.
  • the event object acquires different target state information when the game client executes different target event objects according to the target instruction.
  • Step S1102 Update the processing result of each layer of data combination according to different target state information, and obtain an update processing result of each layer of data combination.
  • step S1102 of the present application after acquiring different target state information when the game client executes different target event objects according to the target instruction, the processing result of each layer data combination is updated according to different target state information, Get the update processing result of each layer of data combination. After obtaining the update processing result of each layer of data combination, the update processing result of the multi-layer data combination is integrated to obtain an update target instruction.
  • the embodiment obtains different target state information when the game client executes different target event objects according to the target instruction; updates the processing result of each layer data combination according to different target state information, and obtains an update processing result of each layer data combination. Improve the efficiency of data processing.
  • Real-time games generally have complex game rules, dynamic scenes, behavioral uncertainty, incomplete information, short decision time, and probability success. Faced with such huge decision space and real-time demand for decision-making, how to formulate, select and execute strategies is the most important game intelligence system. Ask the question.
  • the use of multiple deep learning networks has proven to be more capable of decision making, but not directly applicable to real-time games, where Deep Learning is through the use of multiple complex structures or non-
  • the neural network algorithm of the linear transformation processing layer has better high-level abstraction ability than the shallow neural network.
  • the real-time game is a game type in which the game process is performed in an instant rather than a turn-based game, compared to a turn-based game such as Go or Chess.
  • This embodiment proposes a layered solution that spreads policy choices to multiple levels, so that large flat data can be decentralized, thereby reducing the latitude of the state space and using different algorithms. This embodiment can make tradeoffs between decision time and simulation depth to meet changing game scenarios.
  • This embodiment simulates the decision process of human players, and divides the whole intelligent system into three modules: decision selection, decision making, and feedback tuning, so that the system can cope with the complex and varied scenes of real-time games.
  • decision selection considering the decision depth of game players, the appropriate data samples and algorithms will be selected for decision learning according to the top-down abstraction level, thus reducing the complexity of the operation.
  • micro level the execution of decisions is performed using some quick and simple algorithms that can feed back results without too much decision making.
  • This embodiment can be applied to the human-machine battle mode of the real-time game, and can provide a more anthropomorphic artificial intelligence character to optimize the player's experience.
  • FIG. 12 is a schematic diagram of an interaction process in a game process according to an embodiment of the present invention. As shown in FIG. 12, this embodiment separates policy selection and policy execution.
  • the decision layer includes policy layer 1, policy layer 2 to policy layer n, has a large depth, can simulate a player game decision path, and the policy execution focuses on execution efficiency. , do not make too many decisions, through feedback tuning.
  • the game data design has a lot of latitude and cannot be directly used as training data, and needs to be marked according to predetermined rules.
  • the main method is to mark the entire sample according to a sequence of features of the input sample and a preset logical configuration.
  • 13 is a flow diagram of another method of tagging sample data based on a plurality of sample sequences of sample data, in accordance with an embodiment of the present invention. As shown in FIG. 13, the method for marking sample data according to a plurality of sample sequences of sample data includes the following steps:
  • step S1301 the priority of the sample sequence is separated according to the game characteristics, and the sample sequence order is obtained.
  • the first step is to prioritize the sample sequence according to the game characteristics.
  • step S1302 different sequences are marked in the order of the sample sequence.
  • different sequences are marked according to the sequence of the sample sequences, and different sequences may be marked in the order of the sample sequence using a preset rule or some sample segmentation algorithms.
  • Step S1303 traversing all the marked frames.
  • step S1304 adjacent identical frames are marked as the same sequence.
  • Step S1305 marking the start end frame of each sequence.
  • the sequence of the sample sequence is obtained, the different sequences are marked according to the sequence of the sample sequence, all the mark frames are traversed, and the adjacent frames are marked as the same sequence, and the start of each sequence is marked. Ending the frame, thereby achieving the purpose of marking the sample data according to multiple sample sequences of sample data.
  • FIG. 14 is a flow chart of another data processing method in accordance with an embodiment of the present invention. As shown in FIG. 14, the method for performing preprocessing on the pair of labeled sample data includes the following steps:
  • Step S1401 extracting state information of the current disk surface in the sample by using a general state function.
  • the state information of the current disk surface is extracted in the sample by a general state function, which is called an S state.
  • step S1402 the status information is assembled into a multi-layer data combination.
  • the state information of the current disk surface is extracted in the sample by the general state function, the state information is assembled into a multi-layer data combination.
  • the role status of the party as the first layer the status of the friendly role is placed on the 2-5 layer according to the strength ranking
  • the enemy is placed on the 6-10 layer according to the damage ranking
  • the outermost layer is Map information and NPC information.
  • Map information and NPC information Other data assembly principles can also be used.
  • Step S1403 in each layer of data combination, the state information of the sample, each character data is mapped into a legal action space according to the game rule state, thereby acquiring event data.
  • each character data u is mapped to the legal action space according to the game rule state in each layer of data combination, thereby acquiring the event data a.
  • Step S1404 generating ⁇ S, u, a> according to each motion sample, and rotating the sample to expand the sample size.
  • the state information S, the character data u, and the event data a are generated as ⁇ S, u, a> according to each action sample, and the sample is rotated to expand the sample size.
  • Step S1405 adding preset information to the sample information of the sample size.
  • AI Artificial Intelligence
  • the algorithm can refer to AlphaGo to integrate the strategy network and the value network with the Monte Carlo algorithm. Weighted integration of the output of each decision layer yields the final strategy.
  • the strategy can be added to the status assessment function to determine if there is a need to change the strategy under the current disk to cope with the changing game environment.
  • the corresponding disk surface information is returned to update the learning with the common policy selection module, and the algorithm can use the behavior tree.
  • This embodiment proposes a multi-layered intelligent system architecture, which divides the intelligent system into multiple decision-making layers, simulates the multi-layer abstract decision behavior of the player in the actual game, and separates the decision selection and decision execution to cope with real-time.
  • the needs of the game The decision-making layer applies a multi-layer deep learning framework, constructs samples reasonably, marks and processes the sample strategy sequences, and can cope with different decision-making time requirements. Decision execution is simple and efficient, ensuring fast execution of behavior and improving data processing efficiency.
  • the whole system simulates the player's thinking process, which can effectively improve the AI's ability, thus improving the user experience of the game player.
  • the application environment of the embodiment of the present invention may be, but is not limited to, the reference to the application environment in the foregoing embodiment, which is not described in this embodiment.
  • An embodiment of the present invention provides an optional specific application for implementing the above data processing method.
  • the technical solution of the present invention can be applied to a human-machine battle of a real-time game, and can provide a more anthropomorphic artificial intelligence character, thereby optimizing the experience of the game player.
  • the 15 is a flow chart showing a method of game interaction according to an embodiment of the present invention.
  • the game client acquires the state of the current game and transmits the state of the current game to the policy selection server via the network.
  • the policy selection server is a plurality of servers, the strategy is selected through the model, the best action is selected and returned to the game client, and the game client executes the policy according to the best action, and performs disk information and policy feedback.
  • This embodiment simulates the decision process of the game player, and divides the whole intelligent system into three modules: decision selection, decision execution, and feedback tuning, so that the system can cope with the complicated and varied scenes of the real game.
  • this embodiment considers the player's decision depth problem.
  • the appropriate data samples and algorithms are selected for decision learning, thus reducing the computational complexity.
  • the execution of decisions is performed using some quick and simple algorithms that can feed back results without making too many decisions.
  • the game interaction method is implemented to separate the strategy selection from the strategy execution.
  • the decision-making layer has a large depth and can simulate the decision path of the game player.
  • the execution layer focuses on the execution efficiency, does not make too many decisions, and improves the data processing efficiency.
  • the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, by hardware, but in many cases, the former is A better implementation.
  • the technical solution of the present invention which is essential or contributes to the prior art, may be embodied in the form of a software product stored in a storage medium (such as ROM/RAM, disk,
  • the optical disc includes a number of instructions for causing a terminal device (which may be a cell phone, a computer, a server, or a network device, etc.) to perform the methods described in various embodiments of the present invention.
  • the data processing apparatus may include a first acquisition unit 10, a first processing unit 20, a second processing unit 30, and a third processing unit 40.
  • the first obtaining unit 10 is configured to acquire sample data of a game client execution event.
  • the first processing unit 20 is configured to perform pre-processing on the sample data to obtain a multi-layer data combination, wherein each layer of the data combination in the multi-layer data combination corresponds to one target event object in the same target event, and the multi-layer data combination
  • the different layer data combinations in the pair correspond to different target event objects in the same target event, and the target event object is an event object to be executed simultaneously on the game client.
  • the second processing unit 30 is configured to perform processing on each layer of data combination according to a preset processing algorithm to obtain a processing result of each layer of data combination.
  • the third processing unit 40 is configured to perform integration processing on the processing result of each layer of data combination to obtain a target instruction, where the target instruction is used to instruct the game client to execute different layers simultaneously The different target event objects corresponding to the data combination.
  • the first acquiring unit 10, the first processing unit 20, the second processing unit 30, and the third processing unit 40 may be run in the terminal as part of the device, and may be executed by a processor in the terminal.
  • the functions implemented by the above modules may also be terminal devices such as smart phones (such as Android phones, iOS phones, etc.), tablet computers, applause computers, and mobile Internet devices (MID), PAD, and the like.
  • first obtaining unit 10 in this embodiment may be configured to perform step S402 in Embodiment 1 of the present application
  • first processing unit 20 in this embodiment may be configured to perform Embodiment 1 of the present application
  • second processing unit 30 in this embodiment may be configured to perform step S406 in Embodiment 1 of the present application
  • third processing unit 40 in this embodiment may be configured to perform Embodiment 1 of the present application.
  • the data processing apparatus may include a first acquisition unit 10, a first processing unit 20, a second processing unit 30, and a third processing unit 40.
  • the first processing unit 20 includes a marking module 21 and a processing module 22.
  • first obtaining unit 10, the first processing unit 20, the second processing unit 30, and the third processing unit 40 of this embodiment have the same functions as the data processing apparatus of the embodiment shown in FIG. No longer.
  • the marking module 21 is arranged to mark the sample data according to a plurality of sample sequences of the sample data to obtain the labeled sample data.
  • the processing module 22 is configured to perform pre-processing on the labeled sample data to obtain a multi-layer data combination, wherein different layer data combinations in the multi-layer data combination correspond to different processing algorithms and different sample information.
  • the second processing unit 30 is arranged to perform processing on the sample information in each layer of data combination in accordance with a processing algorithm corresponding to each layer of data combination in the multi-layer data combination, to obtain a processing result of each layer of data combination.
  • the foregoing marking module 21 and the processing module 22 may be run in the terminal as part of the device, and the functions implemented by the above module may be performed by a processor in the terminal, and the terminal may also be a smart phone (such as an Android mobile phone). , iOS phones, etc.), tablet computers, applause computers, and mobile Internet devices (MID), PAD and other terminal devices.
  • a smart phone such as an Android mobile phone. , iOS phones, etc.
  • tablet computers tablet computers, applause computers, and mobile Internet devices (MID), PAD and other terminal devices.
  • MID mobile Internet devices
  • the data processing apparatus may include a first acquisition unit 10, a first processing unit 20, a second processing unit 30, and a third processing unit 40.
  • the first processing unit 20 includes: a marking module 21 and a processing module 22, and the marking module 21 includes: a determining submodule 211, a first marking submodule 212, a merging submodule 213 and a second marking submodule 214.
  • first obtaining unit 10 the first processing unit 20, the second processing unit 30, and the third processing unit 40 of the embodiment, the marking module 21 and the processing module 22 and the data processing of the embodiment shown in FIG.
  • the functions in the device are the same and will not be described here.
  • a determination sub-module 211 is arranged to determine the priority of each of the plurality of sample sequences.
  • the first marking sub-module 212 is arranged to sequentially mark each sample sequence by a marking frame in order of priority to obtain a plurality of marking sample sequences.
  • the merging sub-module 213 is configured to combine adjacent ones of the plurality of labeled sample sequences according to the same marked frame to obtain a merged labeled sample sequence.
  • the second marking sub-module 214 is arranged to mark the start frame and the end frame of the merged mark sample sequence to obtain mark sample data.
  • the foregoing determining submodule 211, the first marking submodule 212, the merging submodule 213 and the second marking submodule 214 may be run in the terminal as part of the device, and may be executed by a processor in the terminal.
  • the functions implemented by the above modules may also be terminal devices such as smart phones (such as Android phones, iOS phones, etc.), tablet computers, applause computers, and mobile Internet devices (MID), PAD, and the like.
  • the data processing apparatus may include a first acquisition unit 10, a first processing unit 20, a second processing unit 30, and a third processing unit 40.
  • the first processing unit 20 includes: a marking module 21 and a processing module 22, and the processing module 22 includes an extraction sub-module 221 and an assembly sub-module 222.
  • first obtaining unit 10 the first processing unit 20, the second processing unit 30, and the third processing unit 40 of the embodiment, the marking module 21 and the processing module 22 and the data processing of the embodiment shown in FIG.
  • the role in the device is the same.
  • the extraction sub-module 221 is configured to extract, in the mark sample data, different state information of the current event object by the game client by using a preset state function, where the current event object is an event object currently executed by the game client.
  • the assembly sub-module 222 is arranged to perform assembly on different state information to obtain a multi-layer data combination.
  • the foregoing extraction submodule 221 and the assembly submodule 222 may be run in the terminal as part of the device, and the function implemented by the above module may be performed by a processor in the terminal, and the terminal may also be a smart phone (eg, Terminal devices such as Android phones, iOS phones, etc., tablets, applause computers, and mobile Internet devices (MID), PAD, etc.
  • Terminal devices such as Android phones, iOS phones, etc., tablets, applause computers, and mobile Internet devices (MID), PAD, etc.
  • the data processing apparatus may include a first acquisition unit 10, a first processing unit 20, a second processing unit 30, and a third processing unit 40.
  • the first processing unit 20 includes: a marking module 21 and a processing module 22, the processing module 22 includes: an extraction sub-module 221 and an assembly sub-module 222, the data processing apparatus further includes: a second acquisition unit 50, a mapping unit 60, and a generation Unit 70.
  • first obtaining unit 10 extracts the submodule 221 and the assembly submodule 222.
  • the functions in the data processing apparatus of the embodiment shown in FIG. 19 are the same, and are not described herein again.
  • the second obtaining unit 50 is configured to perform processing on the sample information in each layer of the data combination according to a processing algorithm corresponding to each layer of the data combination in the multi-layer data combination, and obtain the game before obtaining the processing result of each layer of the data combination Role data on the client.
  • the mapping unit 60 is configured to map the state information and the role data to the preset processing model according to the preset mapping system, to obtain event data of the target event.
  • the generating unit 70 is configured to generate sample information based on the status information, the role data, and the event data.
  • the foregoing second obtaining unit 50, the mapping unit 60 and the generating unit 70 may be operated in the terminal as a part of the device, and the function implemented by the above module may be performed by a processor in the terminal, and the terminal may also be Smartphones (such as Android phones, iOS phones, etc.), tablets, applause computers, and mobile Internet devices (MID), PAD and other terminal devices.
  • Smartphones such as Android phones, iOS phones, etc.
  • MID mobile Internet devices
  • the data processing apparatus may include: a first obtaining unit 10, a first processing unit 20, a second processing unit 30, a third processing unit 40, a second obtaining unit 50, a mapping unit 60, and a generating unit 70.
  • the first processing unit 20 includes: a marking module 21 and a processing module 22, and the processing module 22 includes an extraction sub-module 221 and an assembly sub-module 222.
  • the data processing apparatus further includes: a fourth processing unit 80 and an adding unit 90.
  • first obtaining unit 10 the first processing unit 20, the second processing unit 30, the third processing unit 40, the second obtaining unit 50, the mapping unit 60 and the generating unit 70, and the marking module 21 of this embodiment
  • the processing module 22, the extraction sub-module 221 and the assembly sub-module 222 have the same functions as those in the data processing apparatus of the embodiment shown in FIG. 20, and are not described herein again.
  • the fourth processing unit 80 is configured to perform a rotation process on the sample data to expand the number of samples corresponding to the sample data after generating the sample information according to the state information, the character data, and the event data.
  • the adding unit 90 is set to add preset information to the sample information of the sample number.
  • the fourth processing unit 80 and the adding unit 90 may be operated in the terminal as a part of the device, and the function implemented by the foregoing module may be performed by a processor in the terminal, and the terminal may also be a smart phone (eg, Terminal devices such as Android phones, iOS phones, etc., tablets, applause computers, and mobile Internet devices (MID), PAD, etc.
  • Terminal devices such as Android phones, iOS phones, etc., tablets, applause computers, and mobile Internet devices (MID), PAD, etc.
  • the data processing apparatus may include a first acquisition unit 10, a first processing unit 20, a second processing unit 30, and a third processing unit 40.
  • the second processing unit 30 includes: a first processing module 31 and a second processing module 32.
  • first obtaining unit 10, the first processing unit 20, the second processing unit 30, and the third processing unit 40 of the embodiment have the same functions as the data processing apparatus of the embodiment shown in FIG. Let me repeat.
  • the first processing module 31 is configured to perform processing on the sample information in each layer of data combination according to a preset probability model corresponding to each layer of data combination, to obtain an execution probability of the target event of the game client corresponding to each layer of data combination. .
  • the second processing module 32 is configured to perform processing on the sample information in each layer of data combination according to a preset value model corresponding to each layer of data combination, to obtain an execution value of the execution target event of the game client corresponding to each layer of data combination.
  • the third processing unit 40 is configured to perform an integration process on the execution probability corresponding to each layer of data combination and the execution value corresponding to each layer of data combination to obtain a target instruction.
  • first processing module 31 and the second processing module 32 may be run in the terminal as part of the device, and the function implemented by the module may be performed by a processor in the terminal, and the terminal may also be a smart phone. (such as Android phones, iOS phones, etc.), tablets, applause computers and mobile Internet devices (Mobile Internet Devices, MID), PAD and other terminal devices.
  • the data processing apparatus may include a first acquisition unit 10, a first processing unit 20, a second processing unit 30, a second processing unit 30, and a third processing unit 40.
  • the data processing apparatus further includes: a determining unit 100 and an updating unit 110.
  • first obtaining unit 10, the first processing unit 20, the second processing unit 30, and the third processing unit 40 of this embodiment have the same functions as the data processing apparatus of the embodiment shown in FIG. No longer.
  • the determining unit 100 is configured to acquire the processing result of each layer of data combination, and perform integration processing on the processing result of each layer of data combination, and after obtaining the target instruction, determine whether the target instruction needs to be updated according to the preset state evaluation function.
  • the update unit 110 is configured to update the target instruction when it is determined that the target instruction needs to be updated.
  • the foregoing determining unit 100 and the updating unit 110 may be run in the terminal as a part of the device, and the function implemented by the above module may be performed by a processor in the terminal, and the terminal may also be a smart phone (such as an Android mobile phone). , iOS phones, etc.), tablet computers, applause computers, and mobile Internet devices (MID), PAD and other terminal devices.
  • a smart phone such as an Android mobile phone. , iOS phones, etc.
  • tablet computers tablet computers, applause computers, and mobile Internet devices (MID), PAD and other terminal devices.
  • MID mobile Internet devices
  • the embodiment obtains sample data of the execution event of the game client by the first obtaining unit 10, performs pre-processing on the sample data by the first processing unit 20, and obtains multi-layer data combination, and the data combination of each layer in the multi-layer data combination corresponds to A target event object in the same target event, the different layer data combinations in the multi-layer data combination correspond to different target event objects in the same target event, and the target event object is an event object to be executed simultaneously on the game client,
  • the processing unit 30 performs processing on each layer of data combination according to a preset processing algorithm to obtain a processing result of each layer of data combination, and performs processing on the processing result of each layer of data combination by the third processing unit 40 to obtain a target instruction, a target instruction. It is used to instruct the game client to simultaneously execute different target event objects corresponding to different layer data combinations, and solves the technical problem of low data processing efficiency of the related technology, thereby achieving the technical effect of improving the data processing effect.
  • the above-mentioned units and modules are the same as the examples and application scenarios implemented by the corresponding steps, but are not limited to the contents disclosed in the above embodiments. It should be noted that the foregoing module may be implemented in a hardware environment as shown in FIG. 3 as part of the device, and may be implemented by software or by hardware, where the hardware environment includes a network environment.
  • the various functional modules provided by the embodiments of the present application may be run in a mobile terminal, a computer terminal, or the like, or may be stored as part of a storage medium.
  • embodiments of the present invention may provide a terminal, which may be any one of computer terminal groups.
  • the foregoing terminal may also be replaced with a terminal device such as a mobile terminal.
  • the foregoing terminal may be located in at least one of the plurality of network devices of the computer network.
  • a terminal for implementing the data processing method is further provided, wherein the terminal may be a computer terminal, and the computer terminal may be any one of the computer terminal groups.
  • the foregoing computer terminal may also be replaced with a terminal device such as a mobile terminal.
  • the computer terminal may be located in at least one network device of the plurality of network devices of the computer network.
  • FIG. 24 is a structural block diagram of a terminal according to an embodiment of the present invention.
  • the terminal may include one or more (only one shown in the figure) processor 241, a memory 243, and a transmission device 245.
  • the terminal may further include an input/output device 247. .
  • the memory 243 can be configured to store a software program and a module, such as a data processing method and a program instruction/module corresponding to the device in the embodiment of the present invention, and the processor 241 runs the software program and the module stored in the memory 243, thereby Performing various functional applications and data processing, that is, implementing the above data processing method.
  • Memory 243 can include high speed random access memory, and can also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid state memory.
  • memory 243 can further include relative The memory is remotely set by the processor 241, and the remote memory can be connected to the terminal through a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.
  • the transmission device 245 described above is arranged to receive or transmit data via a network, and may also be configured as a data transmission between the processor and the memory. Specific examples of the above network may include a wired network and a wireless network.
  • the transmission device 245 includes a Network Interface Controller (NIC) that can be connected to other network devices and routers via a network cable to communicate with the Internet or a local area network.
  • the transmission device 245 is a Radio Frequency (RF) module for communicating with the Internet wirelessly.
  • NIC Network Interface Controller
  • RF Radio Frequency
  • the memory 243 is set to store an application.
  • the program code that the processor 241 can call the application stored in the memory 243 by the transmission device 245 to execute the method steps of the various optional or preferred embodiments of the above method embodiments includes:
  • each layer of data combination in the multi-layer data combination corresponds to one target event object in the same target event, and different layer data combinations in the multi-layer data combination correspond to the same target Different target event objects in the event, the target event object is an event object to be executed simultaneously on the game client;
  • the processing result of each layer of data combination is integrated to obtain a target instruction, wherein the target instruction is used to instruct the game client to simultaneously execute different target event objects corresponding to different layer data combinations.
  • the processor 241 is further configured to perform the steps of: marking the sample data according to the plurality of sample sequences of the sample data to obtain the labeled sample data; performing pre-processing on the labeled sample data, Obtaining a multi-layer data combination, wherein different layer data combinations in the multi-layer data combination correspond to different processing algorithms and different sample information; and each layer of data combination is processed according to a processing algorithm corresponding to each layer of data combination in the multi-layer data combination The sample information in the process is processed to obtain the processing result of each layer of data combination.
  • the processor 241 is further configured to perform the steps of: determining a priority of each of the plurality of sample sequences; marking each sample sequence by a marker frame in order of priority to obtain a plurality of marker sample sequences; The adjacent labeled sample sequences in the labeled sample sequence are combined according to the same marked frame to obtain a merged labeled sample sequence; the start frame and the end frame of the merged labeled sample sequence are marked to obtain labeled sample data.
  • the processor 241 is further configured to perform the following steps: extracting, by using a preset state function, the game client to execute different state information of the current event object in the mark sample data, where the current event object is an event object currently executed by the game client; Different state information is assembled to obtain a multi-layer data combination.
  • the processor 241 is further configured to perform the step of performing processing on the sample information in each layer of data combination in accordance with a processing algorithm corresponding to each layer of data combination in the multi-layer data combination, to obtain a processing result of each layer of data combination before Obtaining role data on the game client; mapping state information and role data to a preset processing model according to a preset mapping system, obtaining event data of the target event; and generating sample information according to the state information, the role data, and the event data.
  • the processor 241 is further configured to perform the steps of: performing rotation processing on the sample data to expand the number of samples corresponding to the sample data after generating the sample information according to the state information, the character data, and the event data; adding the preset information to the number of samples In the sample information.
  • the processor 241 is further configured to perform the following steps: performing processing on the sample information in each layer of data combination according to a preset probability model corresponding to each layer of data combination, and obtaining a game client execution target event corresponding to each layer of data combination Execution probability; performing processing on the sample information in each layer of data combination according to a preset value model corresponding to each layer of data combination, and obtaining execution value of the execution target event of the game client corresponding to each layer of data combination; The processing result of the layer data combination, and the processing result of each layer of data combination is integrated to obtain the target instruction
  • the method includes: integrating an execution probability corresponding to each layer of data combination and an execution value corresponding to each layer of data combination to obtain a target instruction.
  • the processor 241 is further configured to perform the following steps: acquiring the processing result of each layer of data combination, and performing integration processing on the processing result of each layer of data combination, and after obtaining the target instruction, determining whether the update is needed according to the preset state evaluation function.
  • the target instruction if it is determined that the target instruction needs to be updated, the target instruction is updated.
  • the processor 241 is further configured to perform the following steps: acquiring the processing result of each layer of data combination, and performing integration processing on the processing result of each layer of the data combination, and obtaining the target instruction, obtaining the different execution according to the target instruction in the game client Different target state information of the target event object; update the processing result of each layer of data combination according to different target state information, and obtain an update processing result of each layer of data combination; wherein, the processing result of each layer of data combination is obtained, and The processing result of each layer of data combination is integrated, and obtaining the target instruction includes: obtaining an update processing result of each layer of data combination, and integrating the update processing result of the multi-layer data combination to obtain an update target instruction.
  • a solution of a data processing method is provided.
  • Each layer of data combination corresponds to one target event object in the same target event, and different layer data combinations in the multi-layer data combination correspond to different target event objects in the same target event, and the target event object is to be simultaneously on the game client.
  • the executed event object the different layer data combination corresponds to different processing algorithms and different sample information; according to the processing algorithm corresponding to each layer of data combination in the multi-layer data combination, the sample information in each layer of data combination is processed, and each is obtained.
  • Different target event objects reached Data processing result of each layer of the multilayer composition to integrate the data combination process, the object to obtain the target instruction, thereby improving the data processing efficiency achieved technical effect, Furthermore, the technical problem of low data processing efficiency of the related art is solved.
  • the terminal may be a smart phone (such as an Android mobile phone, an iOS mobile phone, etc.), a tablet computer, a palm computer, and a mobile Internet device (MID). Terminal equipment such as PAD.
  • Fig. 24 does not limit the structure of the above electronic device.
  • the terminal may also include more or fewer components (such as a network interface, display device, etc.) than shown in FIG. 24, or have a different configuration than that shown in FIG.
  • Embodiments of the present invention also provide a storage medium.
  • the foregoing storage medium may store program code, where the program code is used to perform the steps in the data processing method provided by the foregoing method embodiment.
  • the foregoing storage medium may be located in any one of the computer terminal groups in the computer network, or in any one of the mobile terminal groups.
  • the storage medium is arranged to store program code for performing the following steps:
  • each layer of data combination in the multi-layer data combination corresponds to one target event object in the same target event, and different layer data combinations in the multi-layer data combination correspond to the same target Different target event objects in the event, target events
  • the object is an event object to be executed simultaneously on the game client;
  • the processing result of each layer of data combination is integrated to obtain a target instruction, wherein the target instruction is used to instruct the game client to simultaneously execute different target event objects corresponding to different layer data combinations.
  • the storage medium is further configured to store program code for performing the following steps: marking the sample data according to a plurality of sample sequences of the sample data to obtain labeled sample data; performing pre-processing on the labeled sample data to obtain multiple layers Data combination, wherein different layer data combinations in the multi-layer data combination correspond to different processing algorithms and different sample information; samples in each layer of data combination according to a processing algorithm corresponding to each layer of data combination in the multi-layer data combination The information is processed to obtain the processing result of each layer of data combination.
  • the storage medium is further configured to store program code for performing the steps of: determining a priority of each of the plurality of sample sequences; marking each sample sequence by marker frames in order of priority a plurality of labeled sample sequences; combining adjacent labeled sample sequences of the plurality of labeled sample sequences according to the same marked frame to obtain a merged labeled sample sequence; marking the start frame and the end frame of the merged labeled sample sequence to obtain a mark sample.
  • the storage medium is further configured to store program code for performing the following steps: extracting, by the preset state function, the game client to execute different state information of the current event object in the mark sample data, the current event object being a game client The currently executing event object; performing assembly on different state information to obtain a multi-layer data combination.
  • the storage medium is further configured to store program code for performing processing of performing sample processing on the sample information in each layer of data combination in accordance with a processing algorithm corresponding to each layer of data combination in the combination of the plurality of layers of data
  • the character data on the game client is obtained; the state information and the role data are mapped to the preset processing model according to the preset mapping system, and event data of the target event is obtained; according to the state information, the role data, and Event data Generate sample information.
  • the storage medium is further configured to store program code for performing the following steps: after generating the sample information according to the state information, the role data, and the event data, performing rotation processing on the sample data to expand the number of samples corresponding to the sample data; Add preset information to the sample information of the sample size.
  • the storage medium is further configured to store program code for performing the following steps: performing processing on the sample information in each layer of data combination according to a preset probability model corresponding to each layer of data combination, and obtaining a combination with each layer of data
  • the corresponding game client executes the execution probability of the target event; performs processing on the sample information in each layer of data combination according to the preset value model corresponding to each layer of data combination, and obtains a game client execution target event corresponding to each layer of data combination Execution value; wherein, the processing result of each layer of data combination is obtained, and the processing result of each layer of data combination is integrated, and the target instruction is obtained: the execution probability corresponding to each layer of data combination and the data combination corresponding to each layer The execution value is integrated and the target instruction is obtained.
  • the storage medium is further configured to store program code for performing the following steps: obtaining a processing result of each layer of data combination, and performing integration processing on the processing result of each layer of the data combination, after obtaining the target instruction, according to the pre- Let the state evaluation function determine whether it is necessary to update the target instruction; if it is determined that the target instruction needs to be updated, the target instruction is updated.
  • the storage medium is further configured to store program code for performing the following steps: acquiring the processing result of each layer of the data combination, and integrating the processing result of each layer of the data combination to obtain the target instruction, obtaining the The game client performs different target state information when different target event objects are executed according to the target instruction; updates the processing result of each layer data combination according to different target state information, and obtains an update processing result of each layer of data combination; wherein, each layer is acquired
  • the processing result of the data combination, and the processing result of each layer of data combination is integrated, and the target instruction includes: obtaining the update processing result of each layer data combination, and integrating the update processing result of the multi-layer data combination to obtain an update.
  • Target instruction includes: obtaining the update processing result of each layer data combination, and integrating the update processing result of the multi-layer data combination to obtain an update.
  • the foregoing storage medium may include, but not limited to, a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, and a magnetic
  • ROM read-only memory
  • RAM random access memory
  • mobile hard disk a magnetic
  • magnetic A variety of media that can store program code, such as a disc or a disc.
  • the integrated unit in the above embodiment if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in the above-described computer readable storage medium.
  • the technical solution of the present invention may contribute to the prior art or all or part of the technical solution may be embodied in the form of a software product stored in a storage medium.
  • a number of instructions are included to cause one or more computer devices (which may be a personal computer, server or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention.
  • the disclosed client may be implemented in other manners.
  • the device embodiments described above are merely illustrative.
  • the division of the unit is only a logical function division.
  • multiple units or components may be combined or may be Integrate into another system, or some features can be ignored or not executed.
  • the mutual coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interface, unit or module, and may be electrical or otherwise.
  • the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of the embodiment.
  • each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.
  • the above integrated unit can be implemented in the form of hardware or in the form of a software functional unit.
  • the sample data of the execution event of the game client is acquired; the pre-processing is performed on the sample data to obtain a multi-layer data combination, and each layer of the data combination in the multi-layer data combination corresponds to one target in the same target event.
  • the event object, the different layer data combination in the multi-layer data combination corresponds to different target event objects in the same target event, and the target event object is an event object to be executed simultaneously on the game client; the data combination of each layer is performed according to a preset processing algorithm.
  • the purpose of the instruction is to achieve the technical effect of improving the efficiency of data processing, thereby solving the technical problem of low data processing efficiency of the related art.

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Software Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Human Computer Interaction (AREA)
  • Processing Or Creating Images (AREA)
  • Machine Translation (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

一种数据处理方法、装置和存储介质，所述方法包括：获取游戏客户端执行事件的样本数据（S402）；对样本数据执行预处理，得到多层数据组合（S404），其中，多层数据组合中的每层数据组合对应同一目标事件中的一种目标事件对象，多层数据组合中的不同层数据组合对应同一目标事件中不同的目标事件对象，所述目标事件对象为在所述游戏客户端上待同时执行的事件对象；按照预设处理算法对每层数据组合执行处理，得到每层数据组合的处理结果（S406）；对每层数据组合的处理结果进行整合处理，得到目标指令（S408）。所述方法提高了数据处理效率。

Description

数据处理方法、装置和存储介质
本申请要求于2016年09月21日提交中国专利局、优先权号为2016108388048、发明名称为“数据处理方法和装置”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本发明实施例涉及数据处理领域，具体而言，涉及一种数据处理方法、装置和存储介质。
背景技术
目前，对回合制事件的数据处理规则相对简单，用于对事件执行的对象对事件拥有全局事件信息，并且轮流行动，对事件具有较长的决策时间，事件行为立刻实行，通过反馈进行确定，例如，回合制事件为回合制游戏，其规则相对简单，游戏玩家拥有全局信息，轮流行动，有较长的决策时间，行为可以立刻实行，也可以通过反馈确定，例如，围棋。
而对实时事件的数据处理规则复杂，用于对事件执行的对象只有部分事件信息，同时行动，对事件的决策时间较短，事件行为需要时间执行并且具有一定的成功概率，因此，对实时事件的数据处理与对回合制事件的数据处理存在差异。
在回合制事件中，围棋智能程序阿尔法狗(AlphaGo)是利用深度学习训练策略网络和价值网络，并用蒙特卡洛树整合来实现高水平围棋走子的算法。图1是根据相关技术中的一种AlphaGo算法的决策网络和价值网络的结构示意图。如图1所示，AlphaGo算法采用深度学习训练策略网络和价值网络。人类专家侧(Human expert positions)通过分类(Classification)推出策略(Rollout Policy)，将人类专家侧的策略网络(SL Network)通过 策略算法(Policy Gradient)传输至人工智能侧(Self-play Positions)。在人工智能侧，通过人工智能侧的策略网络(RL Network)和价值网络(Value Network)进行训练，得到数据(Data)，其中，人类专家一侧的策略网络和人工智能侧的策略网络统称为策略网络(Policy Network)，策略网络和价值网络通过一定的算法公式进行训练，通过蒙地卡洛树搜索(Monte Carlo Tree Search，简称为MCTS)算法进行实现。
图2是根据相关技术中的一种蒙地卡洛树搜索算法的示意图。如图2所示，通过策略网络选择落子概率、对落子样本进行扩展、通过价值网络评估当前走子收益、反馈当前走子收益的结果，通过蒙地卡洛树搜索算法对策略网络选择的落子概率和价值网络评估的当前走子的收益进行整合仿真，并最终根据当前盘面选择最佳落子位置。
实时事件的数据处理复杂度远高于上述回合制事件的数据处理复杂度，由于回合制事件的数据处理和实时事件的数据处理存在较多的差异，使得算法的两层网络的结合方式无法满足实时事件宏观决策需求满足，更无法满足微观操作层次，无法满足实时事件智能系统的需要，数据处理效率低。
针对上述的数据处理效率低的问题，目前尚未提出有效的解决方案。
发明内容
本发明实施例提供了一种数据处理方法、装置和存储介质，以至少解决相关技术的数据处理效率低的技术问题。
根据本发明实施例的一个方面，提供了一种数据处理方法。该数据处理方法包括：获取游戏客户端执行事件的样本数据；对样本数据执行预处理，得到多层数据组合，其中，多层数据组合中的每层数据组合对应同一目标事件中的一种目标事件对象，多层数据组合中的不同层数据组合对应同一目标事件中不同的目标事件对象；按照预设处理算法对每层数据组合执行处理，得到每层数据组合的处理结果；对每层数据组合的处理结果进 行整合处理，得到目标指令，其中，目标指令用于指示游戏客户端同时执行不同层数据组合对应的不同的目标事件对象。
根据本发明实施例的另一方面，还提供了一种数据处理装置。该数据处理装置包括：第一获取单元，被设置为获取游戏客户端执行事件的样本数据；第一处理单元，被设置为对样本数据执行预处理，得到多层数据组合，其中，多层数据组合中的每层数据组合对应同一目标事件中的一种目标事件对象，多层数据组合中的不同层数据组合对应同一目标事件中不同的目标事件对象，目标事件对象为在游戏客户端上待同时执行的事件对象；第二处理单元，被设置为按照预设处理算法对每层数据组合执行处理，得到每层数据组合的处理结果；第三处理单元，被设置为对每层数据组合的处理结果进行整合处理，得到目标指令，其中，目标指令用于指示游戏客户端同时执行不同层数据组合对应的不同的目标事件对象。
根据本发明实施例的另一方面，还提供了一种终端，其中，终端被设置为执行程序代码，程序代码用于执行本发明实施例的数据处理方法中的步骤。
根据本发明实施例的另一方面，还提供了一种存储介质，其中，存储介质被设置为存储程序代码，程序代码用于执行本发明实施例的数据处理方法中的步骤。
在本发明实施例中，通过获取游戏客户端执行事件的样本数据；对样本数据执行预处理，得到多层数据组合，多层数据组合中的每层数据组合对应同一目标事件中的一种目标事件对象，多层数据组合中的不同层数据组合对应同一目标事件中不同的目标事件对象，目标事件对象为在游戏客户端上待同时执行的事件对象；按照预设处理算法对每层数据组合执行处理，得到每层数据组合的处理结果；对每层数据组合的处理结果进行整合处理，得到目标指令，达到了对多层数据组合中的每层数据组合的处理结果进行整合处理，得到目标指令的目的，从而实现了提高数据处理效率的技术效果，进而解决了相关技术的数据处理效率低的技术问题。
附图说明
此处所说明的附图用来提供对本发明的进一步理解，构成本申请的一部分，本发明的示意性实施例及其说明用于解释本发明，并不构成对本发明的不当限定。在附图中：
图1是根据相关技术中的一种AlphaGo算法的决策网络和价值网络的结构示意图；
图2是根据相关技术中的一种蒙地卡洛树搜索算法的示意图；
图3是根据本发明实施例的一种数据处理方法的硬件环境的示意图；
图4是根据本发明实施例的一种数据处理方法的流程图；
图5是根据本发明实施例的一种根据样本数据的多个样本序列对样本数据进行标记的方法的流程图；
图6是根据本发明实施例的一种对标记样本数据执行预处理的方法的流程图；
图7是根据本发明实施例的另一种数据处理方法的流程图；
图8是根据本发明实施例的另一种数据处理方法的流程图；
图9是根据本发明实施例的一种按照与多层数据组合中的每层数据组合对应的处理算法对每层数据组合中的样本信息执行处理的方法的流程图；
图10是根据本发明实施例的另一种数据处理方法的流程图；
图11是根据本发明实施例的另一种数据处理方法的流程图；
图12是根据本发明实施例的一种游戏过程中的交互过程的示意图；
图13是根据本发明实施例的另一种根据样本数据的多个样本序列对样本数据进行标记的方法的流程图；
图14是根据本发明实施例的另一种数据处理方法的流程图；
图15是根据本发明实施例的一种游戏交互的方法的流程示意图；
图16是根据本发明实施例的一种数据处理装置的示意图；
图17是根据本发明实施例的另一种数据处理装置的示意图；
图18是根据本发明实施例的另一种数据处理装置的示意图；
图19是根据本发明实施例的另一种数据处理装置的示意图；
图20是根据本发明实施例的另一种数据处理装置的示意图；
图21是根据本发明实施例的另一种数据处理装置的示意图；
图22是根据本发明实施例的另一种数据处理装置的示意图；
图23是根据本发明实施例的另一种数据处理装置的示意图；以及
图24是根据本发明实施例的一种终端的结构框图。
具体实施方式
为了使本技术领域的人员更好地理解本发明方案，下面将结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本发明一部分的实施例，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都应当属于本发明保护的范围。
需要说明的是，本发明的说明书和权利要求书及上述附图中的术语“第一”、“第二”等是用于区别类似的对象，而不必用于描述特定的顺序或先后次序。应该理解这样使用的数据在适当情况下可以互换，以便这里描述的本发明的实施例能够以除了在这里图示或描述的那些以外的顺序实施。此外，术语“包括”和“具有”以及他们的任何变形，意图在于覆盖不排他的包含，例如，包含了一系列步骤或单元的过程、方法、系统、产品或设备不必限于清楚地列出的那些步骤或单元，而是可包括没有清楚地列出 的或对于这些过程、方法、产品或设备固有的其它步骤或单元。
根据本发明实施例，提供了一种数据处理方法的实施例。
可选地，在本实施例中，上述数据处理方法可以应用于如图3所示的由服务器302和终端304所构成的硬件环境中。图3是根据本发明实施例的一种数据处理方法的硬件环境的示意图。如图3所示，服务器302通过网络与终端304进行连接，上述网络包括但不限于：广域网、城域网或局域网，终端304并不限定于PC、手机、平板电脑等。本发明实施例的数据处理方法可以由服务器302来执行，也可以由终端304来执行，还可以是由服务器302和终端304共同执行。其中，终端304执行本发明实施例的数据处理方法也可以是由安装在其上的客户端来执行。
图4是根据本发明实施例的一种数据处理方法的流程图。如图4所示，该数据处理方法可以包括以下步骤：
步骤S402，获取游戏客户端执行事件的样本数据。
在本申请上述步骤S402提供的技术方案中，获取游戏客户端执行事件的样本数据。
游戏客户端用于执行事件，比如，执行实时游戏(Real Time Game)中的人机对战模式中的事件，该实时游戏区别于回合制游戏。
该实施例的游戏客户端在执行事件的过程中会产生数据，该数据可以为游戏数据，样本数据为游戏客户端在执行事件的过程中，实际观测或调查的一部分的数据，可以为随机抽取，具有足够数量，且可以反映游戏客户端执行事件的总体情况的数据。该实施例也可以通过多层深度学习框架对游戏客户端在执行事件的过程中的数据进行合理构建，从而得到样本数据。
可选地，获取游戏客户端执行事件的样本数据，该样本数据也即输入样本，可以为游戏样本，该游戏样本包括多个样本序列，多个样本序列具有不同的优先级，其中，不同优先级的样本序列可以包括相同数据帧。
步骤S404，对样本数据执行预处理，得到多层数据组合。
在本申请上述步骤S404提供的技术方案中，对样本数据执行预处理，得到多层数据组合，其中，多层数据组合中的每层数据组合对应同一目标事件中的一种目标事件对象，多层数据组合中的不同层数据组合对应同一目标事件中不同的目标事件对象，目标事件对象为在游戏客户端上待同时执行的事件对象。
由于游戏数据的设计维度众多，无法直接用作训练数据，在获取游戏客户端执行事件的样本数据之后，获取样本数据的多个样本序列，根据多个样本序列对样本数据进行标记，得到标记样本数据，可以通过预设标记逻辑对样本数据进行标记。可选地，根据输入样本的一个特征序列，根据预设的逻辑配置标记整个样本，进而得到标记样本数据。
由于数据处理规则复杂，会出现多个样本序列包括相同数据帧的情况，也即，同一数据帧属于多个样本序列的情况，可以根据事件的特性分出多个样本序列的优先级。按照样本序列的优先级顺序标记不同的样本序列，可以使用预设规则或者预设样本分割算法对样本序列进行标记，得到标记帧。遍历所有的标记帧，将相邻相同的标记帧标记为一个样本序列，标记各样本序列的起始帧和结束帧，进而得到标记样本序数据。
在根据样本数据的多个样本序列对样本数据进行标记，得到标记样本数据之后，对标记样本数据执行预处理，得到多层数据组合。该实施例的多层数据组合可以由通用的状态函数在标记的样本数据中提取的状态信息组装而成，比如，可以使用通用的状态函数在标记样本数据中提取当前盘面的状态信息，将状态信息组装成多层数据组合。目标事件对象为在游戏客户端上待同时执行的事件对象，可以对应本方角色状态、友方角色状态、敌方伤害力信息、地图信息和非玩家控制角色(Non Player Controlled Character，简称为NPC)信息等。
该实施例的多层数据组合中的每层数据组合对应同一目标事件中的一种目标事件对象，多层数据组合中的不同层数据组合对应同一目标事件 中不同的目标事件对象，也即，多层数据组合中的每层数据组合与同一目标事件中的目标事件对象具有一一对应关系，比如，目标事件为对战事件，目标事件对象包括待同时执行的A事件对象、B事件对象、C事件对象，其中，A事件对象、B事件对象、C事件对象不同，则第一层数据组合对应对战事件中的A事件对象，第二层数据组合对应对战事件中的B事件对象，第三层数据组合对应对战事件中的C事件对象。再比如，待同时执行的事件对象包括本方角色、友方角色、敌方、非玩家控制角色，每层数据组合对应本方角色、友方角色、敌方、非玩家控制角色等。该实施例的多层数据组合中的不同层数据组合对应同一目标事件中不同的目标事件对象，比如，在一个双方各五人的实时游戏中，将本方角色状态作为第一层，将友方角色状态按照实力排名作为第二层至第五层，也即，第二层至第五层由于实力排名的不同将友方角色区分开，敌方按照伤害力排名作为第六层至第十层，也即，第六层至第十层由于伤害力排名的不同将敌方区分开，地图信息和NPC信息作为最外层，其它数据组装原则也适用，此处不做限制。将每层数据组合中的样本数据的状态信息，每个角色数据根据游戏规则状态映射至某一个合法的动作空间中，从而获得事件数据。将每层数据组合中的状态信息、每个角色数据、每个事件数据组成每层数据组合中的样本信息。对样本数据执行旋转处理以扩展样本数据对应的样本数量，其它用户信息也可以加入样本信息中，比如，将执行事件的失误率、操作频次加入样本信息中，从而便于训练，实现了对标记样本数据执行预处理，得到多层数据组合的目的。
步骤S406，按照预设处理算法对每层数据组合执行处理，得到每层数据组合的处理结果。
在本申请上述步骤S406提供的技术方案中，按照预设处理算法对每层数据组合执行处理，得到每层数据组合的处理结果。
多层数据组合中的每层数据组合具有对应的预设处理算法，该预设处理算法可以为在每层数据组合中，按照样本信息学习当前状态信息下的事 件执行概率的概率模型和事件的执行价值的价值模型，比如，根据样本中当前盘面的状态信息、角色信息、角色根据游戏规则状态映射到某一个合法的动作空间中对应的数值，学习当前状态下动作执行概率模型和动作的价值模型，具体算法可以为AlphaGo整合策略网络和价值网络的蒙特卡洛树搜索算法。其中，蒙特卡洛树搜索算法是一种用于决策的启发式搜索算法，通过扩展搜索树和仿真来选择受益最大的行为从而做出最优决策，得到每层数据组合的处理结果，该处理结果用于决策层输出以进行整合处理，为按照预设处理算法对每层数据组合执行处理得到的结果，该处理结果包括执行概率和执行价值，可以按照与每层数据组合对应的预设概率模型、预设价值模型对每层数据组合中的样本信息执行处理，得到与每层数据组合对应的游戏客户端执行目标事件的执行概率和执行价值，比如，预设概率模型为动作执行概率模型，预设价值模型为动作的价值模型，根据当前状态下动作执行概率模型和动作的价值模型对每层数据组合中的样本信息执行处理，得到与每层数据组合对应的游戏客户端执行目标事件的执行概率和执行价值。
步骤S408，对每层数据组合的处理结果进行整合处理，得到目标指令。
在本申请上述步骤S408提供的技术方案中，对每层数据组合的处理结果进行整合处理，得到目标指令，其中，目标指令用于指示游戏客户端同时执行不同层数据组合对应的不同的目标事件对象。
在按照与对每层数据组合中的样本信息执行处理，得到每层数据组合的处理结果之后，对每层数据组合的处理结果进行加权整合，得到目标指令，该目标指令也即最终策略，用于指示游戏客户端同时执行不同层数据组合对应的不同的目标事件对象。该目标指令可以加入状态评估函数，以确定当前盘面下是否需要改变目标指令，从而满足多变的事件执行环境。
可选地，在对每层数据组合的处理结果进行整合处理，得到目标指令之后，执行该目标指令，也即，执行最终的策略。在目标指令执行的过程 中，可以加入状态评估函数，以确定当前盘面下是否需要改变策略，以应对多变的游戏环境。
可选地，在执行目标指令的过程中，可以显示游戏客户端当前的盘面信息，该当前的盘面信息也即对目标指令进行指令的执行结果，根据预设状态评估函数和当前的执行结果判断是否需要更新目标指令，可以通过行为树(Behavior Tree)判断是否需要更新目标指令。其中，行为树为一种图形化的模型语言，在游戏中用来描述不同行为的执行条件和方式，进而保证行为的快速执行，以提升游戏玩家的体验。
通过上述步骤S402至步骤S408，通过获取游戏客户端执行事件的样本数据；对样本数据执行预处理，得到多层数据组合，多层数据组合中的每层数据组合对应同一目标事件中的一种目标事件对象，多层数据组合中的不同层数据组合对应同一目标事件中不同的目标事件对象，目标事件对象为在游戏客户端上待同时执行的事件对象；按照与多层数据组合中的每层数据组合对应的处理算法对每层数据组合中的样本信息执行处理，得到每层数据组合的处理结果；对每层数据组合的处理结果进行整合处理，得到目标指令，其中，目标指令用于指示游戏客户端同时执行不同层数据组合对应的不同的目标事件对象，可以解决相关技术的数据处理效率低的技术问题，进而达到提高数据处理效率的技术效果。
作为一种可选的实施方式，步骤S404，对样本数据执行预处理，得到多层数据组合包括：根据样本数据的多个样本序列对样本数据进行标记，得到标记样本数据；对标记样本数据执行预处理，得到多层数据组合，其中，多层数据组合中的不同层数据组合对应不同的处理算法和不同的样本信息。这样多层数据组合各自使用不同的算法，可以在决策时长和仿真深度上做折衷，以满足多变的游戏场景，可以应对不同决策时长的要求，这样决策执行使用简单高效，从而保证了行为的快速执行。
在按照预设处理算法对每层数据组合执行处理，得到每层数据组合的处理结果时，可以按照与多层数据组合中的每层数据组合对应的处理算法 对每层数据组合中的样本信息执行处理，得到每层数据组合的处理结果。
作为一种可选的实施方式，在步骤S404中，根据样本数据的多个样本序列对样本数据进行标记，得到标记样本数据包括：按照多个样本序列中每个样本序列的优先级依次通过标记帧对每个样本序列进行标记，得到多个标记样本序列，再将多个标记样本序列中相邻的标记样本序列按照相同的标记帧进行合并得到的合并标记样本序列中的起始帧和结束帧进行标记，得到标记样本数据。
图5是根据本发明实施例的一种根据样本数据的多个样本序列对样本数据进行标记的方法的流程图。如图5所示，该根据样本数据的多个样本序列对样本数据进行标记的方法包括以下步骤：
步骤S501，确定多个样本序列中每个样本序列的优先级。
在本申请上述步骤S501提供的技术方案中，确定多个样本序列中每个样本序列的优先级。样本数据包括多个样本序列，多个样本序列中每个样本序列具有优先级，由于事件执行规则复杂，会出现同一数据帧属于多个样本序列的情况，在获取游戏客户端执行事件的样本数据之后，确定多个样本序列中每个样本序列的优先级，进而根据每个样本序列的优先级得出多个样本序列的排列顺序。
步骤S502，按照优先级依次通过标记帧对每个样本序列进行标记，得到多个标记样本序列。
在本申请上述步骤S502提供的技术方案中，按照优先级依次通过标记帧对每个样本序列进行标记，得到多个标记样本序列。在确定多个样本序列中每个样本序列的优先级之后，可以按照多个样本序列的排列顺序通过标记帧标记不同的样本序列。可选地，通过预设规则或者预设样本分割算法来通过标记帧按照多个样本序列的排列顺序标记不同的样本序列，得到多个标记样本序列。
步骤S503，将多个标记样本序列中相邻的标记样本序列按照相同的 标记帧进行合并，得到合并标记样本序列。
在本申请上述步骤S503提供的技术方案中，在按照优先级依次通过标记帧对每个样本序列进行标记，得到多个标记样本序列之后，将多个标记样本序列中相邻的标记样本序列按照相同的标记帧进行合并，得到合并标记样本序列。可以遍历多个标记样本序列的标记帧，将相邻相同标记帧的样本序列标记为同一个序列，得到合并标记样本序列。
步骤S504，对合并标记样本序列的起始帧和结束帧进行标记，得到标记样本数据。
在本申请上述步骤S504提供的技术方案中，在将多个标记样本序列中相邻的标记样本序列按照相同的标记帧进行合并，得到合并标记样本序列之后，对合并标记样本序列的起始帧和结束帧进行标记，得到标记样本数据。可以在将多个标记样本序列中相邻的标记样本序列按照相同的标记帧进行合并，得到合并标记样本序列之后，标记各合并标记样本序列的起始帧和结束帧，得到标记样本数据。
该实施例通过确定多个样本序列中每个样本序列的优先级；按照优先级依次通过标记帧对每个样本序列进行标记，得到多个标记样本序列；将多个标记样本序列中相邻的标记样本序列按照相同的标记帧进行合并，得到合并标记样本序列；对合并标记样本序列的起始帧和结束帧进行标记，得到标记样本数据，实现了根据样本数据的多个样本序列对样本数据进行标记，得到标记样本数据的目的。
作为一种可选的实施方式，在步骤S404中，对标记样本数据执行预处理，得到多层数据组合包括：对游戏客户端执行当前事件对象的不同的状态信息执行组装，得到多层数据组合。
图6是根据本发明实施例的一种对标记样本数据执行预处理的方法的流程图。如图6所示，该对标记样本数据执行预处理的方法包括以下步骤：
步骤S601，通过预设状态函数在标记样本数据中提取游戏客户端执 行当前事件对象的不同的状态信息。
在本申请上述步骤S601提供的技术方案中，在根据样本数据的多个样本序列对样本数据进行标记，得到标记样本数据之后，通过通用的状态函数在标记样本数据中提取当前盘面的不同的状态信息，该不同的状态信息可以用于指示游戏客户端在实时游戏过程中的本方角色状态、友方角色状态、敌方角色状态等状态信息，当前事件对象为游戏客户端当前执行的事件对象。
步骤S602，对不同的状态信息执行组装，得到多层数据组合。
在本申请上述步骤S602提供的技术方案中，在通过预设状态函数在标记样本数据中提取游戏客户端执行当前事件对象的不同的状态信息之后，将不同的状态信息执行组装，比如，将本方角色状态、友方角色状态、敌方角色状态等状态信息执行组装，得到多层数据组合，该多层数据组合中的每层数据组合对应同一目标事件中的一种目标事件对象，多层数据组合中的不同层数据组合对应同一目标事件中不同的目标事件对象，该目标事件对象为在游戏客户端上待同时执行的事件对象。
举例而言，在一个双方各五人的实时游戏中，本方角色状态作为第一层，友方角色状态按照实力排名放在第二层至第五层，敌方角色状态按照伤害力排名放在第六层至第十层，最外层为地图信息和NPC信息，其它的数据组装方法也可以适用。
该实施例通过预设状态函数在标记样本数据中提取游戏客户端执行当前事件对象的不同的状态信息，当前事件对象为游戏客户端当前执行的事件对象；对不同的状态信息执行组装，得到多层数据组合，实现了对标记样本数据执行预处理，得到多层数据组合的目的，进而提高了数据处理效率。
作为一种可选的实施方式，在按照与多层数据组合中的每层数据组合对应的处理算法对每层数据组合中的样本信息执行处理，得到每层数据组 合的处理结果之前，将状态信息、游戏客户端上的角色数据和目标事件的事件数据生成样本信息。
图7是根据本发明实施例的另一种数据处理方法的流程图。如图7所示，该数据处理方法包括以下步骤：
步骤S701，获取游戏客户端上的角色数据。
在本申请上述步骤S701提供的技术方案中，游戏客户端在执行目标事件时，具有角色数据，该角色数据用于表示执行目标事件的虚拟应用主体的数据。该角色数据包括多个角色数据，每个角色数据对应一个虚拟应用主体的数据，获取游戏客户端上的每个角色数据。
步骤S702，将状态信息和角色数据按照预设映射系映射至预设处理模型，得到目标事件的事件数据。
在本申请上述步骤S702提供的技术方案中，状态信息和角色数据可以通过游戏规则状态映射到合法的动作空间中，该动作空间中具有和状态信息、角色数据相对应的目标事件的事件数据，该事件数据可以为动作数据，从而通过状态信息和角色数据以及合法的动作空间得到目标事件的事件数据。
步骤S703，根据状态信息、角色数据和事件数据生成样本信息。
在本申请上述步骤S703提供的技术方案中，在将状态信息和角色数据按照预设映射系映射至预设处理模型，得到目标事件的事件数据之后，根据状态信息、角色数据和事件数据生成样本信息，该样本信息包括帧信息，每个事件数据对应一个样本信息。比如，将状态信息、角色数据和事件数据以<S，u，a>表示，其中，S用于表示状态信息、u用于表示角色数据，a用于表示事件数据。
该实施例获取游戏客户端上的角色数据；将状态信息和角色数据按照预设映射系映射至预设处理模型，得到目标事件的事件数据；根据状态信息、角色数据和事件数据生成样本信息，进而通过在按照与多层数据组合 中的每层数据组合对应的处理算法对每层数据组合中的样本信息执行处理，得到每层数据组合的处理结果，进而提高了数据处理的效率。
作为一种可选的实施方式，在根据状态信息、角色数据和事件数据生成样本信息之后，添加预设信息至样本数量的样本信息中。
图8是根据本发明实施例的另一种数据处理方法的流程图。如图8所示，该数据处理方法还包括以下步骤：
步骤S801，对样本数据执行旋转处理以扩展样本数据对应的样本数量。
在本申请上述步骤S801提供的技术方案中，在根据状态信息、角色数据和事件数据生成样本信息之后，对样本数据执行旋转处理以扩展样本数据对应的样本数量。
步骤S802，添加预设信息至样本数量的样本信息中。
在本申请上述步骤S802提供的技术方案中，在对样本数据执行旋转处理以扩展样本数据对应的样本数量之后，添加预设信息至样本数量的样本信息中。该预设信息可以为其他用户的信息，比如，失误率，操作频次等信息，将失误率，操作频次等信息加入帧信息中，以便于训练个性化的策略。
作为一种可选的实施方式，在步骤S406中，按照与多层数据组合中的每层数据组合对应的处理算法对每层数据组合中的样本信息执行处理，得到每层数据组合的处理结果包括：按照与每层数据组合对应的预设概率模型、预设价值模型对每层数据组合中的样本信息执行处理，得到与每层数据组合对应的游戏客户端执行目标事件的执行概率和执行价值，并通过与每层数据组合对应的执行概率和与每层数据组合对应的执行价值得到目标指令。
图9是根据本发明实施例的一种按照与多层数据组合中的每层数据组合对应的处理算法对每层数据组合中的样本信息执行处理的方法的流程 图。如图9所示，该方法包括以下步骤：
步骤S901，按照与每层数据组合对应的预设概率模型对每层数据组合中的样本信息执行处理，得到与每层数据组合对应的游戏客户端执行目标事件的执行概率。
在本申请上述步骤S901提供的技术方案中，每层数据组合对应预设概率模型，每层数据组合根据每层数据组合的样本信息学习当前状态下动作执行的预设概率模型，得到与每层数据组合对应的游戏客户端执行目标事件的执行概率。
步骤S902，按照与每层数据组合对应的预设价值模型对每层数据组合中的样本信息执行处理，得到与每层数据组合对应的游戏客户端执行目标事件的执行价值。
在本申请上述步骤S902提供的技术方案中，每层数据组合对应预设价值模型，每层数据组合根据每层数据组合的样本信息学习当前状态下动作执行的预设价值模型，得到与每层数据组合对应的游戏客户端执行目标事件的价值概率。
步骤S903，对与每层数据组合对应的执行概率和与每层数据组合对应的执行价值进行整合处理，得到目标指令。
在本申请上述步骤S903提供的技术方案中，在得到与每层数据组合对应的游戏客户端执行目标事件的执行概率，与每层数据组合对应的游戏客户端执行目标事件的执行价值之后，对与每层数据组合对应的执行概率和与每层数据组合对应的执行价值进行整合处理，得到目标指令，输出最终策略，在策略进行中，加入状态评估函数，从而确定当前盘面下是否需要改变策略，应对多变的事件执行环境。
该实施例通过按照与每层数据组合对应的预设概率模型对每层数据组合中的样本信息执行处理，得到与每层数据组合对应的游戏客户端执行目标事件的执行概率；按照与每层数据组合对应的预设价值模型对每层数 据组合中的样本信息执行处理，得到与每层数据组合对应的游戏客户端执行目标事件的执行价值，实现了按照与多层数据组合中的每层数据组合对应的处理算法对每层数据组合中的样本信息执行处理，得到每层数据组合的处理结果，对与每层数据组合对应的执行概率和与每层数据组合对应的执行价值进行整合处理，得到目标指令，提高了数据处理效率。
作为一种可选的实施方式，在对每层数据组合的处理结果进行整合处理，得到目标指令之后，在需要更新目标指令的情况下，对目标指令进行更新。
图10是根据本发明实施例的另一种数据处理方法的流程图。如图10所示，该数据处理方法包括以下步骤：
步骤S1001，根据预设状态评估函数判断是否需要更新目标指令。
在本申请上述步骤S1001提供的技术方案中，在获取每层数据组合的处理结果，并对每层数据组合的处理结果进行整合处理，得到目标指令之后，根据目标指令执行目标事件，返回对应的盘面信息，根据预设状态评估函数判断是否需要更新目标指令，可以通过行为树判断是否需要更新目标指令。
步骤S1002，如果判断出需要更新目标指令，对目标指令进行更新。
在本申请上述步骤S1002提供的技术方案中，在根据预设状态评估函数判断是否需要更新目标指令之后，如果判断出需要更新目标指令，对目标指令进行更新，从而应对多变的事件处理环境。
该实施例通过在对每层数据组合的处理结果进行整合处理，得到目标指令之后，根据预设状态评估函数判断是否需要更新目标指令；如果判断出需要更新目标指令，对目标指令进行更新，提高了数据处理效率。
作为一种可选的实施方式，在对每层数据组合的处理结果进行整合处理，得到目标指令之后，根据目标指令执行不同的目标事件对象时的不同的目标状态信息更新每层数据组合的处理结果，得到每层数据组合的更新 处理结果，通过对多层数据组合的更新处理结果进行整合处理，得到更新目标指令。
图11是根据本发明实施例的另一种数据处理方法的流程图。如图11所示，该数据处理方法包括以下步骤：
步骤S1101，获取在游戏客户端根据目标指令执行不同的目标事件对象时的不同的目标状态信息。
在本申请上述步骤S1101提供的技术方案中，在获取每层数据组合的处理结果，并对每层数据组合的处理结果进行整合处理，得到目标指令之后，游戏客户端根据目标指令执行不同的目标事件对象，获取在游戏客户端根据目标指令执行不同的目标事件对象时的不同的目标状态信息。
步骤S1102，根据不同的目标状态信息更新每层数据组合的处理结果，得到每层数据组合的更新处理结果。
在本申请上述步骤S1102提供的技术方案中，获取在游戏客户端根据目标指令执行不同的目标事件对象时的不同的目标状态信息之后，根据不同的目标状态信息更新每层数据组合的处理结果，得到每层数据组合的更新处理结果。在得到每层数据组合的更新处理结果之后，对多层数据组合的更新处理结果进行整合处理，得到更新目标指令。
该实施例通过获取在游戏客户端根据目标指令执行不同的目标事件对象时的不同的目标状态信息；根据不同的目标状态信息更新每层数据组合的处理结果，得到每层数据组合的更新处理结果，提高了数据处理的效率。
下面结合优选的实施例对本发明的技术方案进行说明。具体以游戏智能系统进行举例说明。
实时游戏一般都有复杂的游戏规则、多变的动态场景、行为不确定、信息不完全、决策时间短、概率成功等特点。面对如此巨大的决策空间和决策的实时需求，如何制定、选择和执行策略是游戏智能系统面对的最主 要问题。在回合制游戏上，利用多个深度学习网络的方法已经被证明具有较强的决策能力，但是无法直接应用在实时游戏上，其中，深度学习(Deep Learning)是通过使用多个复杂结构或非线性变换处理层的神经网络算法，较浅层神经网络有更好的高层抽象能力，实时游戏是一种游戏过程即时进行而不是采用回合制的游戏类型，相对于围棋、象棋等回合制游戏。
该实施例提出一种分层解决方法，把策略选择分散到多个层次，从而使庞大的扁平化数据能够被分散学习，进而降低了状态空间的纬度，并能各自使用不同算法。该实施例可以在决策时长和仿真深度上做折衷，以满足多变的游戏场景。
该实施例模拟人类玩家的决策过程，将整个智能系统分为决策选择、决策实行、反馈调优三个模块，使得系统能够应对实时游戏复杂多变的场景。在宏观层面，考虑了游戏玩家的决策深度问题，将根据自上而下的抽象层次不同，选取合适的数据样本和算法进行决策学习，从而降低运算的复杂度。在微观层面，决策的执行使用某些快速简单的算法执行，能够反馈结果，不需进行过多的决策考量。
该实施例可以应用在实时游戏的人机对战模式中，可以提供更加拟人化的人工智能角色，优化玩家的体验。
图12是根据本发明实施例的一种游戏过程中的交互过程的示意图。如图12所示，该实施例将策略选择和策略执行分离，决策层包括策略层1、策略层2至策略层n，具有较大的深度，能够模拟玩家游戏决策路径，策略执行关注执行效率，不作过多决策，通过反馈调优。
游戏数据设计纬度众多，无法直接用作训练数据，需要根据预定规则进行标记。主要方法是根据输入样本的一个特征序列、预设的逻辑配置标记整个样本。图13是根据本发明实施例的另一种根据样本数据的多个样本序列对样本数据进行标记的方法的流程图。如图13所示，该根据样本数据的多个样本序列对样本数据进行标记的方法包括以下步骤：
步骤S1301，按照游戏特性分出样本序列的优先级，得到样本序列顺序。
因为游戏规则复杂，会出现同一帧属于多种序列的情况，所以第一步需要根据游戏特性分出样本序列的优先级。
步骤S1302，按照样本序列顺序标记不同序列。
在按照游戏特性分出样本序列的优先级，得到样本序列顺序之后，按照样本序列顺序标记不同序列，可以使用预设规则或者某些样本分割算法实现按照样本序列顺序标记不同序列。
步骤S1303，遍历所有标记帧。
在按照样本序列顺序标记不同序列之后，遍历所有标记帧，标记后向第一个已标记帧的序列。
步骤S1304，将相邻相同帧标记为同一个序列。
在遍历所有帧的过程中，将相邻相同帧标记为同一个序列。
步骤S1305，标记各序列的起始结束帧。
在将相邻相同帧标记为同一个序列之后，标记各序列的起始帧和结束帧。
该实施例通过按照游戏特性分出样本序列的优先级，得到样本序列顺序，按照样本序列顺序标记不同序列，遍历所有标记帧，将相邻相同帧标记为同一个序列，标记各序列的起始结束帧，从而实现了根据样本数据的多个样本序列对样本数据进行标记的目的。
图14是根据本发明实施例的另一种数据处理方法的流程图。如图14所示，该对标记样本数据执行预处理方法包括以下步骤：
步骤S1401，通过通用状态函数在样本中提取当前盘面的状态信息。
通过通用状态函数在样本中提取当前盘面的状态信息，称为S状态。
步骤S1402，将状态信息组装成多层数据组合。
在通过通用状态函数在样本中提取当前盘面的状态信息之后，将状态信息组装成多层数据组合。假设一个双方各5人的实时游戏，本方角色状态作为第一层，友方角色状态按照实力排名放在2-5层，敌方按伤害力排名放在6-10层，最外层为地图信息和NPC信息。其他数据组装原则也可使用。
步骤S1403，在每层数据组合将样本的状态信息，每个角色数据根据游戏规则状态映射到合法的动作空间中，从而获取事件数据。
在将状态信息组装成多层数据组合之后，在每层数据组合将样本的状态信息，每个角色数据u根据游戏规则状态映射到合法的动作空间中，从而获取事件数据a。
步骤S1404，根据每个动作样本生成<S，u，a>，对样本进行旋转以扩充样本数量。
在获取事件数据a之后，根据每个动作样本将状态信息S、角色数据u和事件数据a生成<S，u，a>，对样本进行旋转以扩充样本数量。
步骤S1405，添加预设信息至样本数量的样本信息中。
添加例如失误率的其他用户信息，操作频次也可加入帧信息中，以便于训练个性化人工智能(Artificial Intelligence，简称为AI)。
在每个决策层根据样本的<S，u，a>信息，学习当前状态下动作执行概率模型和动作的价值模型，算法可以参照AlphaGo用蒙地卡洛树算法整合策略网络和价值网络。加权整合各个决策层的输出，得出最终策略。策略进行中可加入状态评估函数，以确定当前盘面下是否需要改变策略，从而应对多变的游戏环境。
在策略执行时，返回对应的盘面信息以共策略选择模块更新学习，算法可以使用行为树。
该实施例提出了一种多层的智能系统架构，将智能系统分成多个决策层的构建思路，模拟玩家在实际游戏中的多层抽象决策行为，将决策选择和决策执行分离，以应对实时游戏的需求。决策层应用多层深度学习框架，合理构建样本，进行样本策略序列的标记与处理，并且可以应对不同决策时长的要求。决策执行使用简单高效，可以保证行为的快速执行，提高了数据处理效率。整个系统模拟玩家的思考过程，能够有效地提升AI的能力，从而提升了游戏玩家的用户体验。
本发明实施例的应用环境可以但不限于参照上述实施例中的应用环境，本实施例中对此不再赘述。本发明实施例提供了用于实施上述数据处理方法的一种可选的具体应用。
本发明的技术方案可以应用在实时游戏的人机对战中，可以提供更加拟人化的人工智能角色，从而优化游戏玩家的体验。
图15是根据本发明实施例的一种游戏交互的方法的流程示意图。如图15所示，游戏客户端获取当前游戏的状态，将当前游戏的状态通过网络发送到策略选择服务器上。策略选择服务器为多个服务器，通过模型进行策略选择，选择最佳的动作并返回给游戏客户端，游戏客户端根据最佳的动作执行策略，并进行盘面信息、策略反馈。
该实施例模拟游戏玩家的决策过程，将整个智能系统分为决策选择、决策实行、反馈调优三个模块，使得系统能够应对实时游戏复杂多变的场景。在宏观层面，该实施例考虑了玩家决策深度问题，根据自上而下的抽象层次不同，选取合适的数据样本和算法进行决策学习，从而降低运算的复杂度。在微观层面，决策的执行使用某些快速简单的算法执行，能够反馈结果，不需进行过多的决策。实现了游戏交互方法将策略选择和策略执行分开，决策层具有较大的深度，能够模拟游戏玩家决策路径，执行层侧重于执行效率，不做过多决策，提高了数据处理效率。
需要说明的是，对于前述的各方法实施例，为了简单描述，故将其都表述为一系列的动作组合，但是本领域技术人员应该知悉，本发明并不受 所描述的动作顺序的限制，因为依据本发明，某些步骤可以采用其他顺序或者同时进行。其次，本领域技术人员也应该知悉，说明书中所描述的实施例均属于优选实施例，所涉及的动作和模块并不一定是本发明所必须的。
通过以上的实施方式的描述，本领域的技术人员可以清楚地了解到根据上述实施例的方法可借助软件加必需的通用硬件平台的方式来实现，当然也可以通过硬件，但很多情况下前者是更佳的实施方式。基于这样的理解，本发明的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来，该计算机软件产品存储在一个存储介质(如ROM/RAM、磁碟、光盘)中，包括若干指令用以使得一台终端设备(可以是手机，计算机，服务器，或者网络设备等)执行本发明各个实施例所述的方法。
根据本发明实施例，还提供了一种用于实施上述数据处理方法的数据处理装置。图16是根据本发明实施例的一种数据处理装置的示意图。如图16所示，该数据处理装置可以包括：第一获取单元10，第一处理单元20，第二处理单元30和第三处理单元40。
第一获取单元10，被设置为获取游戏客户端执行事件的样本数据。
第一处理单元20，被设置为对样本数据执行预处理，得到多层数据组合，其中，多层数据组合中的每层数据组合对应同一目标事件中的一种目标事件对象，多层数据组合中的不同层数据组合对应同一目标事件中不同的目标事件对象，目标事件对象为在游戏客户端上待同时执行的事件对象。
第二处理单元30，被设置为按照预设处理算法对每层数据组合执行处理，得到每层数据组合的处理结果。
第三处理单元40，被设置为对每层数据组合的处理结果进行整合处理，得到目标指令，其中，目标指令用于指示游戏客户端同时执行不同层 数据组合对应的不同的目标事件对象。
此处需要说明的是，上述第一获取单元10，第一处理单元20，第二处理单元30和第三处理单元40可以作为装置的一部分运行在终端中，可以通过终端中的处理器来执行上述模块实现的功能，终端也可以是智能手机(如Android手机、iOS手机等)、平板电脑、掌声电脑以及移动互联网设备(Mobile Internet Devices，MID)、PAD等终端设备。
需要说明的是，该实施例中的第一获取单元10可以被设置为执行本申请实施例1中的步骤S402，该实施例中的第一处理单元20可以被设置为执行本申请实施例1中的步骤S404，该实施例中的第二处理单元30可以被设置为执行本申请实施例1中的步骤S406，该实施例中的第三处理单元40可以被设置为执行本申请实施例1中的步骤S408。
图17是根据本发明实施例的另一种数据处理装置的示意图。如图17所示，该数据处理装置可以包括：第一获取单元10，第一处理单元20，第二处理单元30和第三处理单元40。其中，第一处理单元20包括：标记模块21和处理模块22。
需要说明的是，该实施例的第一获取单元10，第一处理单元20，第二处理单元30和第三处理单元40与图16所示实施例的数据处理装置中的作用相同，此处不再赘述。
标记模块21，被设置为根据样本数据的多个样本序列对样本数据进行标记，得到标记样本数据。
处理模块22，被设置为对标记样本数据执行预处理，得到多层数据组合，其中，多层数据组合中的不同层数据组合对应不同的处理算法和不同的样本信息。
第二处理单元30被设置为按照与多层数据组合中的每层数据组合对应的处理算法对每层数据组合中的样本信息执行处理，得到每层数据组合的处理结果。
此处需要说明的是，上述标记模块21和处理模块22可以作为装置的一部分运行在终端中，可以通过终端中的处理器来执行上述模块实现的功能，终端也可以是智能手机(如Android手机、iOS手机等)、平板电脑、掌声电脑以及移动互联网设备(Mobile Internet Devices，MID)、PAD等终端设备。
图18是根据本发明实施例的另一种数据处理装置的示意图。如图18所示，该数据处理装置可以包括：第一获取单元10，第一处理单元20，第二处理单元30和第三处理单元40。其中，第一处理单元20包括：标记模块21和处理模块22，标记模块21包括：确定子模块211，第一标记子模块212，合并子模块213和第二标记子模块214。
需要说明的是，该实施例的第一获取单元10，第一处理单元20，第二处理单元30和第三处理单元40，标记模块21和处理模块22与图17所示实施例的数据处理装置中的作用相同，此处不再赘述。
确定子模块211，被设置为确定多个样本序列中每个样本序列的优先级。
第一标记子模块212，被设置为按照优先级依次通过标记帧对每个样本序列进行标记，得到多个标记样本序列。
合并子模块213，被设置为将多个标记样本序列中相邻的标记样本序列按照相同的标记帧进行合并，得到合并标记样本序列。
第二标记子模块214，被设置为对合并标记样本序列的起始帧和结束帧进行标记，得到标记样本数据。
此处需要说明的是，上述确定子模块211，第一标记子模块212，合并子模块213和第二标记子模块214可以作为装置的一部分运行在终端中，可以通过终端中的处理器来执行上述模块实现的功能，终端也可以是智能手机(如Android手机、iOS手机等)、平板电脑、掌声电脑以及移动互联网设备(Mobile Internet Devices，MID)、PAD等终端设备。
图19是根据本发明实施例的另一种数据处理装置的示意图。如图19所示，该数据处理装置可以包括：第一获取单元10，第一处理单元20，第二处理单元30和第三处理单元40。其中，第一处理单元20包括：标记模块21和处理模块22，处理模块22包括：提取子模块221和组装子模块222。
需要说明的是，该实施例的第一获取单元10，第一处理单元20，第二处理单元30和第三处理单元40，标记模块21和处理模块22与图17所示实施例的数据处理装置中的作用相同。
提取子模块221，被设置为通过预设状态函数在标记样本数据中提取游戏客户端执行当前事件对象的不同的状态信息，当前事件对象为游戏客户端当前执行的事件对象。
组装子模块222，被设置为对不同的状态信息执行组装，得到多层数据组合。
此处需要说明的是，上述提取子模块221和组装子模块222可以作为装置的一部分运行在终端中，可以通过终端中的处理器来执行上述模块实现的功能，终端也可以是智能手机(如Android手机、iOS手机等)、平板电脑、掌声电脑以及移动互联网设备(Mobile Internet Devices，MID)、PAD等终端设备。
图20是根据本发明实施例的另一种数据处理装置的示意图。如图20所示，该数据处理装置可以包括：第一获取单元10，第一处理单元20，第二处理单元30和第三处理单元40。其中，第一处理单元20包括：标记模块21和处理模块22，处理模块22包括：提取子模块221和组装子模块222，该数据处理装置还包括：第二获取单元50，映射单元60和生成单元70。
需要说明的是，该实施例的第一获取单元10，第一处理单元20，第二处理单元30和第三处理单元40，提取子模块221和组装子模块222与 图19所示实施例的数据处理装置中的作用相同，此处不再赘述。
第二获取单元50，被设置为在按照与多层数据组合中的每层数据组合对应的处理算法对每层数据组合中的样本信息执行处理，得到每层数据组合的处理结果之前，获取游戏客户端上的角色数据。
映射单元60，被设置为将状态信息和角色数据按照预设映射系映射至预设处理模型，得到目标事件的事件数据。
生成单元70，被设置为根据状态信息、角色数据和事件数据生成样本信息。
此处需要说明的是，上述第二获取单元50，映射单元60和生成单元70可以作为装置的一部分运行在终端中，可以通过终端中的处理器来执行上述模块实现的功能，终端也可以是智能手机(如Android手机、iOS手机等)、平板电脑、掌声电脑以及移动互联网设备(Mobile Internet Devices，MID)、PAD等终端设备。
图21是根据本发明实施例的另一种数据处理装置的示意图。如图21所示，该数据处理装置可以包括：第一获取单元10，第一处理单元20，第二处理单元30，第三处理单元40，第二获取单元50，映射单元60和生成单元70。其中，第一处理单元20包括：标记模块21和处理模块22，处理模块22包括：提取子模块221和组装子模块222。该数据处理装置还包括：第四处理单元80和添加单元90。
需要说明的是，该实施例的第一获取单元10，第一处理单元20，第二处理单元30，第三处理单元40，第二获取单元50，映射单元60和生成单元70，标记模块21和处理模块22，提取子模块221和组装子模块222与图20所示实施例的数据处理装置中的作用相同，此处不再赘述。
第四处理单元80，被设置为在根据状态信息、角色数据和事件数据生成样本信息之后，对样本数据执行旋转处理以扩展样本数据对应的样本数量。
添加单元90，被设置为添加预设信息至样本数量的样本信息中。
此处需要说明的是，上述第四处理单元80和添加单元90可以作为装置的一部分运行在终端中，可以通过终端中的处理器来执行上述模块实现的功能，终端也可以是智能手机(如Android手机、iOS手机等)、平板电脑、掌声电脑以及移动互联网设备(Mobile Internet Devices，MID)、PAD等终端设备。
图22是根据本发明实施例的另一种数据处理装置的示意图。如图22所示，该数据处理装置可以包括：第一获取单元10，第一处理单元20，第二处理单元30和第三处理单元40。其中，第二处理单元30包括：第一处理模块31和第二处理模块32。
需要说明的是，该实施例的第一获取单元10，第一处理单元20，第二处理单元30第三处理单元40与图16所示实施例的数据处理装置中的作用相同，此处不再赘述。
第一处理模块31，被设置为按照与每层数据组合对应的预设概率模型对每层数据组合中的样本信息执行处理，得到与每层数据组合对应的游戏客户端执行目标事件的执行概率。
第二处理模块32，被设置为按照与每层数据组合对应的预设价值模型对每层数据组合中的样本信息执行处理，得到与每层数据组合对应的游戏客户端执行目标事件的执行价值。
第三处理单元40被设置为对与每层数据组合对应的执行概率和与每层数据组合对应的执行价值进行整合处理，得到目标指令。
此处需要说明的是，上述第一处理模块31和第二处理模块32可以作为装置的一部分运行在终端中，可以通过终端中的处理器来执行上述模块实现的功能，终端也可以是智能手机(如Android手机、iOS手机等)、平板电脑、掌声电脑以及移动互联网设备(Mobile Internet Devices，MID)、PAD等终端设备。
图23是根据本发明实施例的另一种数据处理装置的示意图。如图23所示，该数据处理装置可以包括：第一获取单元10，第一处理单元20，第二处理单元30，第二处理单元30和第三处理单元40。该数据处理装置还包括：判断单元100和更新单元110。
需要说明的是，该实施例的第一获取单元10，第一处理单元20，第二处理单元30和第三处理单元40与图16所示实施例的数据处理装置中的作用相同，此处不再赘述。
判断单元100，被设置为在获取每层数据组合的处理结果，并对每层数据组合的处理结果进行整合处理，得到目标指令之后，根据预设状态评估函数判断是否需要更新目标指令。
更新单元110，被设置为在判断出需要更新目标指令，对目标指令进行更新。
此处需要说明的是，上述判断单元100和更新单元110可以作为装置的一部分运行在终端中，可以通过终端中的处理器来执行上述模块实现的功能，终端也可以是智能手机(如Android手机、iOS手机等)、平板电脑、掌声电脑以及移动互联网设备(Mobile Internet Devices，MID)、PAD等终端设备。
本发明实施例通过第一获取单元10获取游戏客户端执行事件的样本数据，通过第一处理单元20对样本数据执行预处理，得到多层数据组合，多层数据组合中的每层数据组合对应同一目标事件中的一种目标事件对象，多层数据组合中的不同层数据组合对应同一目标事件中不同的目标事件对象，目标事件对象为在游戏客户端上待同时执行的事件对象，通过第二处理单元30按照预设处理算法对每层数据组合执行处理，得到每层数据组合的处理结果，通过第三处理单元40对每层数据组合的处理结果进行整合处理，得到目标指令，目标指令用于指示游戏客户端同时执行不同层数据组合对应的不同的目标事件对象，解决了相关技术的数据处理效率低的技术问题，进而达到提高数据处理效果的技术效果。
此处需要说明的是，上述单元和模块与对应的步骤所实现的示例和应用场景相同，但不限于上述实施例所公开的内容。需要说明的是，上述模块作为装置的一部分可以运行在如图3所示的硬件环境中，可以通过软件实现，也可以通过硬件实现，其中，硬件环境包括网络环境。
本申请实施例所提供的各个功能模块可以在移动终端、计算机终端或者类似的运算装置中运行，也可以作为存储介质的一部分进行存储。
由此，本发明的实施例可以提供一种终端，该终端可以是计算机终端群中的任意一个计算机终端设备。可选地，在本实施例中，上述终端也可以替换为移动终端等终端设备。
可选地，在本实施例中，上述终端可以位于计算机网络的多个网络设备中的至少一个网络设备。
根据本发明实施例，还提供了一种用于实施上述数据处理方法的终端，其中，终端就可以为计算机终端，该计算机终端可以是计算机终端群中的任意一个计算机终端设备。可选地，在本实施例中，上述计算机终端也可以替换为移动终端等终端设备。
可选地，在本实施例中，上述计算机终端可以位于计算机网络的多个网络设备中的至少一个网络设备。
图24是根据本发明实施例的一种终端的结构框图。如图24所示，该终端可以包括：一个或多个(图中仅示出一个)处理器241、存储器243、以及传输装置245，如图24所示，该终端还可以包括输入输出设备247。
其中，存储器243可被设置为存储软件程序以及模块，如本发明实施例中的数据处理方法和装置对应的程序指令/模块，处理器241通过运行存储在存储器243内的软件程序以及模块，从而执行各种功能应用以及数据处理，即实现上述的数据处理方法。存储器243可包括高速随机存储器，还可以包括非易失性存储器，如一个或者多个磁性存储装置、闪存、或者其他非易失性固态存储器。在一些实例中，存储器243可进一步包括相对 于处理器241远程设置的存储器，这些远程存储器可以通过网络连接至终端。上述网络的实例包括但不限于互联网、企业内部网、局域网、移动通信网及其组合。
上述的传输装置245被设置为经由一个网络接收或者发送数据，还可以被设置为处理器与存储器之间的数据传输。上述的网络具体实例可包括有线网络及无线网络。在一个实例中，传输装置245包括一个网络适配器(Network Interface Controller，NIC)，其可通过网线与其他网络设备与路由器相连从而可与互联网或局域网进行通讯。在一个实例中，传输装置245为射频(Radio Frequency，RF)模块，其用于通过无线方式与互联网进行通讯。
其中，具体地，存储器243被设置为存储应用程序。
处理器241可以通过传输装置245调用存储器243存储的应用程序，以执行上述方法实施例中的各个可选或优选实施例的方法步骤的程序代码，包括：
获取游戏客户端执行事件的样本数据；
对样本数据执行预处理，得到多层数据组合，其中，多层数据组合中的每层数据组合对应同一目标事件中的一种目标事件对象，多层数据组合中的不同层数据组合对应同一目标事件中不同的目标事件对象，目标事件对象为在游戏客户端上待同时执行的事件对象；
按照预设处理算法对每层数据组合中的样本信息执行处理，得到每层数据组合的处理结果；
对每层数据组合的处理结果进行整合处理，得到目标指令，其中，目标指令用于指示游戏客户端同时执行不同层数据组合对应的不同的目标事件对象。
处理器241还被设置为执行下述步骤：根据样本数据的多个样本序列对样本数据进行标记，得到标记样本数据；对标记样本数据执行预处理， 得到多层数据组合，其中，多层数据组合中的不同层数据组合对应不同的处理算法和不同的样本信息；按照与多层数据组合中的每层数据组合对应的处理算法对每层数据组合中的样本信息执行处理，得到每层数据组合的处理结果。
处理器241还被设置为执行下述步骤：确定多个样本序列中每个样本序列的优先级；按照优先级依次通过标记帧对每个样本序列进行标记，得到多个标记样本序列；将多个标记样本序列中相邻的标记样本序列按照相同的标记帧进行合并，得到合并标记样本序列；对合并标记样本序列的起始帧和结束帧进行标记，得到标记样本数据。
处理器241还被设置为执行下述步骤：通过预设状态函数在标记样本数据中提取游戏客户端执行当前事件对象的不同的状态信息，当前事件对象为游戏客户端当前执行的事件对象；对不同的状态信息执行组装，得到多层数据组合。
处理器241还被设置为执行下述步骤：在按照与多层数据组合中的每层数据组合对应的处理算法对每层数据组合中的样本信息执行处理，得到每层数据组合的处理结果之前，获取游戏客户端上的角色数据；将状态信息和角色数据按照预设映射系映射至预设处理模型，得到目标事件的事件数据；根据状态信息、角色数据和事件数据生成样本信息。
处理器241还被设置为执行下述步骤：在根据状态信息、角色数据和事件数据生成样本信息之后，对样本数据执行旋转处理以扩展样本数据对应的样本数量；添加预设信息至样本数量的样本信息中。
处理器241还被设置为执行下述步骤：按照与每层数据组合对应的预设概率模型对每层数据组合中的样本信息执行处理，得到与每层数据组合对应的游戏客户端执行目标事件的执行概率；按照与每层数据组合对应的预设价值模型对每层数据组合中的样本信息执行处理，得到与每层数据组合对应的游戏客户端执行目标事件的执行价值；其中，获取每层数据组合的处理结果，并对每层数据组合的处理结果进行整合处理，得到目标指令 包括：对与每层数据组合对应的执行概率和与每层数据组合对应的执行价值进行整合处理，得到目标指令。
处理器241还被设置为执行下述步骤：在获取每层数据组合的处理结果，并对每层数据组合的处理结果进行整合处理，得到目标指令之后，根据预设状态评估函数判断是否需要更新目标指令；如果判断出需要更新目标指令，对目标指令进行更新。
处理器241还被设置为执行下述步骤：在获取每层数据组合的处理结果，并对每层数据组合的处理结果进行整合处理，得到目标指令之后，获取在游戏客户端根据目标指令执行不同的目标事件对象时的不同的目标状态信息；根据不同的目标状态信息更新每层数据组合的处理结果，得到每层数据组合的更新处理结果；其中，获取每层数据组合的处理结果，并对每层数据组合的处理结果进行整合处理，得到目标指令包括：获取每层数据组合的更新处理结果，并对多层数据组合的更新处理结果进行整合处理，得到更新目标指令。
采用本发明实施例，提供了一种数据处理方法的方案。通过获取游戏客户端执行事件的样本数据；根据样本数据的多个样本序列对样本数据进行标记，得到标记样本数据；对标记样本数据执行预处理，得到多层数据组合，其中，多层数据组合中的每层数据组合对应同一目标事件中的一种目标事件对象，多层数据组合中的不同层数据组合对应同一目标事件中不同的目标事件对象，目标事件对象为在游戏客户端上待同时执行的事件对象，不同层数据组合对应不同的处理算法和不同的样本信息；按照与多层数据组合中的每层数据组合对应的处理算法对每层数据组合中的样本信息执行处理，得到每层数据组合的处理结果；获取每层数据组合的处理结果，并对每层数据组合的处理结果进行整合处理，得到目标指令，其中，目标指令用于指示游戏客户端同时执行不同层数据组合对应的不同的目标事件对象，达到了对多层数据组合中的每层数据组合的处理结果进行整合处理，得到目标指令的目的，从而实现了提高数据处理效率的技术效果， 进而解决了相关技术的数据处理效率低的技术问题。
可选地，本实施例中的具体示例可以参考上述实施例中所描述的示例，本实施例在此不再赘述。
本领域普通技术人员可以理解，图24所示的结构仅为示意，终端可以是智能手机(如Android手机、iOS手机等)、平板电脑、掌上电脑以及移动互联网设备(Mobile Internet Devices，MID)、PAD等终端设备。图24其并不对上述电子装置的结构造成限定。例如，终端还可包括比图24中所示更多或者更少的组件(如网络接口、显示装置等)，或者具有与图24所示不同的配置。
本领域普通技术人员可以理解上述实施例的各种方法中的全部或部分步骤是可以通过程序来指令终端设备相关的硬件来完成，该程序可以存储于一计算机可读存储介质中，存储介质可以包括：闪存盘、只读存储器(Read-Only Memory，ROM)、随机存取器(Random Access Memory，RAM)、磁盘或光盘等。
本发明的实施例还提供了一种存储介质。可选地，在本实施例中，上述存储介质可以存储程序代码，所述程序代码用于执行上述方法实施例所提供的数据处理方法中的步骤。
可选地，在本实施例中，上述存储介质可以位于计算机网络中计算机终端群中的任意一个计算机终端中，或者位于移动终端群中的任意一个移动终端中。
可选地，在本实施例中，存储介质被设置为存储用于执行以下步骤的程序代码：
获取游戏客户端执行事件的样本数据；
对样本数据执行预处理，得到多层数据组合，其中，多层数据组合中的每层数据组合对应同一目标事件中的一种目标事件对象，多层数据组合中的不同层数据组合对应同一目标事件中不同的目标事件对象，目标事件 对象为在游戏客户端上待同时执行的事件对象；
按照预设处理算法对每层数据组合中的样本信息执行处理，得到每层数据组合的处理结果；
对每层数据组合的处理结果进行整合处理，得到目标指令，其中，目标指令用于指示游戏客户端同时执行不同层数据组合对应的不同的目标事件对象。
可选地，存储介质还被设置为存储用于执行以下步骤的程序代码：根据样本数据的多个样本序列对样本数据进行标记，得到标记样本数据；对标记样本数据执行预处理，得到多层数据组合，其中，多层数据组合中的不同层数据组合对应不同的处理算法和不同的样本信息；按照与多层数据组合中的每层数据组合对应的处理算法对每层数据组合中的样本信息执行处理，得到每层数据组合的处理结果。
可选地，存储介质还被设置为存储用于执行以下步骤的程序代码：确定多个样本序列中每个样本序列的优先级；按照优先级依次通过标记帧对每个样本序列进行标记，得到多个标记样本序列；将多个标记样本序列中相邻的标记样本序列按照相同的标记帧进行合并，得到合并标记样本序列；对合并标记样本序列的起始帧和结束帧进行标记，得到标记样本数据。
可选地，存储介质还被设置为存储用于执行以下步骤的程序代码：通过预设状态函数在标记样本数据中提取游戏客户端执行当前事件对象的不同的状态信息，当前事件对象为游戏客户端当前执行的事件对象；对不同的状态信息执行组装，得到多层数据组合。
可选地，存储介质还被设置为存储用于执行以下步骤的程序代码：在按照与多层数据组合中的每层数据组合对应的处理算法对每层数据组合中的样本信息执行处理，得到每层数据组合的处理结果之前，获取游戏客户端上的角色数据；将状态信息和角色数据按照预设映射系映射至预设处理模型，得到目标事件的事件数据；根据状态信息、角色数据和事件数据 生成样本信息。
可选地，存储介质还被设置为存储用于执行以下步骤的程序代码：在根据状态信息、角色数据和事件数据生成样本信息之后，对样本数据执行旋转处理以扩展样本数据对应的样本数量；添加预设信息至样本数量的样本信息中。
可选地，存储介质还被设置为存储用于执行以下步骤的程序代码：按照与每层数据组合对应的预设概率模型对每层数据组合中的样本信息执行处理，得到与每层数据组合对应的游戏客户端执行目标事件的执行概率；按照与每层数据组合对应的预设价值模型对每层数据组合中的样本信息执行处理，得到与每层数据组合对应的游戏客户端执行目标事件的执行价值；其中，获取每层数据组合的处理结果，并对每层数据组合的处理结果进行整合处理，得到目标指令包括：对与每层数据组合对应的执行概率和与每层数据组合对应的执行价值进行整合处理，得到目标指令。
可选地，存储介质还被设置为存储用于执行以下步骤的程序代码：在获取每层数据组合的处理结果，并对每层数据组合的处理结果进行整合处理，得到目标指令之后，根据预设状态评估函数判断是否需要更新目标指令；如果判断出需要更新目标指令，对目标指令进行更新。
可选地，存储介质还被设置为存储用于执行以下步骤的程序代码：在获取每层数据组合的处理结果，并对每层数据组合的处理结果进行整合处理，得到目标指令之后，获取在游戏客户端根据目标指令执行不同的目标事件对象时的不同的目标状态信息；根据不同的目标状态信息更新每层数据组合的处理结果，得到每层数据组合的更新处理结果；其中，获取每层数据组合的处理结果，并对每层数据组合的处理结果进行整合处理，得到目标指令包括：获取每层数据组合的更新处理结果，并对多层数据组合的更新处理结果进行整合处理，得到更新目标指令。
可选地，本实施例中的具体示例可以参考上述实施例中所描述的示例，本实施例在此不再赘述。
可选地，在本实施例中，上述存储介质可以包括但不限于：U盘、只读存储器(Read-Only Memory，ROM)、随机存取存储器(Random Access Memory，RAM)、移动硬盘、磁碟或者光盘等各种可以存储程序代码的介质。
如上参照附图以示例的方式描述了根据本发明的数据处理方法的更新方法、装置和存储介质。但是，本领域技术人员应当理解，对于上述本发明所提出的虚拟应用属性的更新方法、装置和存储介质，还可以在不脱离本发明内容的基础上做出各种改进。因此，本发明的保护范围应当由所附的权利要求书的内容确定。
上述本发明实施例序号仅仅为了描述，不代表实施例的优劣。
上述实施例中的集成的单元如果以软件功能单元的形式实现并作为独立的产品销售或使用时，可以存储在上述计算机可读取的存储介质中。基于这样的理解，本发明的技术方案本质上或者说对现有技术做出贡献的部分或者该技术方案的全部或部分可以以软件产品的形式体现出来，该计算机软件产品存储在存储介质中，包括若干指令用以使得一台或多台计算机设备(可为个人计算机、服务器或者网络设备等)执行本发明各个实施例所述方法的全部或部分步骤。
在本发明的上述实施例中，对各个实施例的描述都各有侧重，某个实施例中没有详述的部分，可以参见其他实施例的相关描述。
在本申请所提供的几个实施例中，应该理解到，所揭露的客户端，可通过其它的方式实现。其中，以上所描述的装置实施例仅仅是示意性的，例如所述单元的划分，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式，例如多个单元或组件可以结合或者可以集成到另一个系统，或一些特征可以忽略，或不执行。另一点，所显示或讨论的相互之间的耦合或直接耦合或通信连接可以是通过一些接口，单元或模块的间接耦合或通信连接，可以是电性或其它的形式。
所述作为分离部件说明的单元可以是或者也可以不是物理上分开的，作为单元显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部单元来实现本实施例方案的目的。
另外，在本发明各个实施例中的各功能单元可以集成在一个处理单元中，也可以是各个单元单独物理存在，也可以两个或两个以上单元集成在一个单元中。上述集成的单元既可以采用硬件的形式实现，也可以采用软件功能单元的形式实现。
以上所述仅是本发明的优选实施方式，应当指出，对于本技术领域的普通技术人员来说，在不脱离本发明原理的前提下，还可以做出若干改进和润饰，这些改进和润饰也应视为本发明的保护范围。
工业实用性
在本发明实施例中，通过获取游戏客户端执行事件的样本数据；对样本数据执行预处理，得到多层数据组合，多层数据组合中的每层数据组合对应同一目标事件中的一种目标事件对象，多层数据组合中的不同层数据组合对应同一目标事件中不同的目标事件对象，目标事件对象为在游戏客户端上待同时执行的事件对象；按照预设处理算法对每层数据组合执行处理，得到每层数据组合的处理结果；对每层数据组合的处理结果进行整合处理，得到目标指令，达到了对多层数据组合中的每层数据组合的处理结果进行整合处理，得到目标指令的目的，从而实现了提高数据处理效率的技术效果，进而解决了相关技术的数据处理效率低的技术问题。

Claims (19)

  1. 一种数据处理方法，包括：
    获取游戏客户端执行事件的样本数据；
    对所述样本数据执行预处理，得到多层数据组合，其中，所述多层数据组合中的每层数据组合对应同一目标事件中的一种目标事件对象，所述多层数据组合中的不同层数据组合对应同一目标事件中不同的目标事件对象，所述目标事件对象为在所述游戏客户端上待同时执行的事件对象；
    按照预设处理算法对所述每层数据组合执行处理，得到所述每层数据组合的处理结果；
    对所述每层数据组合的处理结果进行整合处理，得到目标指令，其中，所述目标指令用于指示所述游戏客户端同时执行所述不同层数据组合对应的不同的目标事件对象。
  2. 根据权利要求1所述的方法，其中，
    对所述样本数据执行预处理，得到所述多层数据组合包括：根据所述样本数据的多个样本序列对所述样本数据进行标记，得到标记样本数据；对所述标记样本数据执行预处理，得到所述多层数据组合，其中，所述多层数据组合中的不同层数据组合对应不同的处理算法和不同的样本信息；
    按照所述预设处理算法对所述每层数据组合执行处理，得到所述每层数据组合的处理结果包括：按照与所述多层数据组合中的每层数据组合对应的处理算法对所述每层数据组合中的样本信息执行处理，得到所述每层数据组合的处理结果。
  3. 根据权利要求2所述的方法，其中，根据所述样本数据的多个样本序列对所述样本数据进行标记，得到所述标记样本数据包括：
    确定所述多个样本序列中每个样本序列的优先级；
    按照所述优先级依次通过标记帧对所述每个样本序列进行标记，得到多个标记样本序列；
    将所述多个标记样本序列中相邻的标记样本序列按照相同的所述标记帧进行合并，得到合并标记样本序列；
    对所述合并标记样本序列的起始帧和结束帧进行标记，得到所述标记样本数据。
  4. 根据权利要求2所述的方法，其中，对所述标记样本数据执行预处理，得到所述多层数据组合包括：
    通过预设状态函数在所述标记样本数据中提取所述游戏客户端执行当前事件对象的不同的状态信息，其中，所述当前事件对象为所述游戏客户端当前执行的事件对象；
    对所述不同的状态信息执行组装，得到所述多层数据组合。
  5. 根据权利要求4所述的方法，其中，在按照与所述多层数据组合中的每层数据组合对应的处理算法对所述每层数据组合中的样本信息执行处理，得到所述每层数据组合的处理结果之前，所述方法还包括：
    获取所述游戏客户端上的角色数据；
    将所述状态信息和所述角色数据按照预设映射系映射至预设处理模型，得到所述目标事件的事件数据；
    根据所述状态信息、所述角色数据和所述事件数据生成所述样本信息。
  6. 根据权利要求5所述的方法，其中，在根据所述状态信息、所述角色数据和所述事件数据生成所述样本信息之后，所述方法还包括：
    对所述样本数据执行旋转处理以扩展所述样本数据对应的样本数量；
    添加预设信息至所述样本数量的样本信息中。
  7. 根据权利要求2所述的方法，其中，按照与所述多层数据组合中的每层数据组合对应的处理算法对所述每层数据组合中的样本信息执行 处理，得到所述每层数据组合的处理结果包括：
    按照与所述每层数据组合对应的预设概率模型对所述每层数据组合中的样本信息执行处理，得到与所述每层数据组合对应的所述游戏客户端执行所述目标事件的执行概率；
    按照与所述每层数据组合对应的预设价值模型对所述每层数据组合中的样本信息执行处理，得到与所述每层数据组合对应的所述游戏客户端执行所述目标事件的执行价值；
    其中，对所述每层数据组合的处理结果进行整合处理，得到所述目标指令包括：对与所述每层数据组合对应的所述执行概率和与所述每层数据组合对应的所述执行价值进行整合处理，得到所述目标指令。
  8. 根据权利要求1所述的方法，其中，在对所述每层数据组合的处理结果进行整合处理，得到所述目标指令之后，所述方法还包括：
    根据预设状态评估函数判断是否需要更新所述目标指令；
    如果判断出需要更新所述目标指令，对所述目标指令进行更新。
  9. 根据权利要求1所述的方法，其中，在对所述每层数据组合的处理结果进行整合处理，得到所述目标指令之后，所述方法还包括：
    获取在所述游戏客户端根据所述目标指令执行所述不同的目标事件对象时的不同的目标状态信息；
    根据所述不同的目标状态信息更新所述每层数据组合的处理结果，得到所述每层数据组合的更新处理结果；
    其中，对所述每层数据组合的处理结果进行整合处理，得到所述目标指令包括：获取所述每层数据组合的所述更新处理结果，并对所述多层数据组合的更新处理结果进行整合处理，得到更新目标指令。
  10. 一种数据处理装置，包括：
    第一获取单元，被设置为获取游戏客户端执行事件的样本数据；
    第一处理单元，被设置为对所述样本数据执行预处理，得到多层 数据组合，其中，所述多层数据组合中的每层数据组合对应同一目标事件中的一种目标事件对象，所述多层数据组合中的不同层数据组合对应同一目标事件中不同的目标事件对象，所述目标事件对象为在所述游戏客户端上待同时执行的事件对象；
    第二处理单元，被设置为按照预设处理算法对所述每层数据组合执行处理，得到所述每层数据组合的处理结果；
    第三处理单元，被设置为对所述每层数据组合的处理结果进行整合处理，得到目标指令，其中，所述目标指令用于指示所述游戏客户端同时执行所述不同层数据组合对应的不同的目标事件对象。
  11. 根据权利要求10所述的装置，其中，
    所述第一处理单元包括：标记模块，被设置为根据所述样本数据的多个样本序列对所述样本数据进行标记，得到标记样本数据；处理模块，被设置为对所述标记样本数据执行预处理，得到所述多层数据组合，其中，所述多层数据组合中的不同层数据组合对应不同的处理算法和不同的样本信息；
    所述第二处理单元被设置为按照与所述多层数据组合中的每层数据组合对应的处理算法对所述每层数据组合中的样本信息执行处理，得到所述每层数据组合的处理结果。
  12. 根据权利要求11所述的装置，其中，所述标记模块包括：
    确定子模块，被设置为确定所述多个样本序列中每个样本序列的优先级；
    第一标记子模块，被设置为按照所述优先级依次通过标记帧对所述每个样本序列进行标记，得到多个标记样本序列；
    合并子模块，被设置为将所述多个标记样本序列中相邻的标记样本序列按照相同的所述标记帧进行合并，得到合并标记样本序列；
    第二标记子模块，被设置为对所述合并标记样本序列的起始帧和结束帧进行标记，得到所述标记样本数据。
  13. 根据权利要求11所述的装置，其中，所述处理模块包括：
    提取子模块，被设置为通过预设状态函数在所述标记样本数据中提取所述游戏客户端执行当前事件对象的不同的状态信息，其中，所述当前事件对象为所述游戏客户端当前执行的事件对象；
    组装子模块，被设置为对所述不同的状态信息执行组装，得到所述多层数据组合。
  14. 根据权利要求13所述的装置，其中，所述装置还包括：
    第二获取单元，被设置为在按照与所述多层数据组合中的每层数据组合对应的处理算法对所述每层数据组合中的样本信息执行处理，得到所述每层数据组合的处理结果之前，获取所述游戏客户端上的角色数据；
    映射单元，被设置为将所述状态信息和所述角色数据按照预设映射系映射至预设处理模型，得到所述目标事件的事件数据；
    生成单元，被设置为根据所述状态信息、所述角色数据和所述事件数据生成所述样本信息。
  15. 根据权利要求14所述的装置，其中，所述装置还包括：
    第四处理单元，被设置为在根据所述状态信息、所述角色数据和所述事件数据生成所述样本信息之后，对所述样本数据执行旋转处理以扩展所述样本数据对应的样本数量；
    添加单元，被设置为添加预设信息至所述样本数量的样本信息中。
  16. 根据权利要求11所述的装置，其中，所述第二处理单元包括：
    第一处理模块，被设置为按照与所述每层数据组合对应的预设概率模型对所述每层数据组合中的样本信息执行处理，得到与所述每层数据组合对应的所述游戏客户端执行所述目标事件的执行概率；
    第二处理模块，被设置为按照与所述每层数据组合对应的预设价值模型对所述每层数据组合中的样本信息执行处理，得到与所述每层 数据组合对应的所述游戏客户端执行所述目标事件的执行价值；
    其中，所述第三处理单元被设置为对与所述每层数据组合对应的所述执行概率和与所述每层数据组合对应的所述执行价值进行整合处理，得到所述目标指令。
  17. 根据权利要求10所述的装置，其中，所述装置还包括：
    判断单元，被设置为在获取所述每层数据组合的所述处理结果，并对所述每层数据组合的处理结果进行整合处理，得到所述目标指令之后，根据预设状态评估函数判断是否需要更新所述目标指令；
    更新单元，被设置为在判断出需要更新所述目标指令，对所述目标指令进行更新。
  18. 一种终端，其中，所述终端被设置为执行程序代码，所述程序代码用于执行所述权利要求1至9中任意一项所述的数据处理方法中的步骤。
  19. 一种存储介质，其中，所述存储介质被设置为存储程序代码，所述程序代码用于执行所述权利要求1至9中任意一项所述的数据处理方法中的步骤。
PCT/CN2017/102702 2016-09-21 2017-09-21 数据处理方法、装置和存储介质 WO2018054330A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
EP17852398.1A EP3518105B1 (en) 2016-09-21 2017-09-21 Method and device for data processing and storage medium
US16/221,842 US11135514B2 (en) 2016-09-21 2018-12-17 Data processing method and apparatus, and storage medium for concurrently executing event characters on a game client

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201610838804.8 2016-09-21
CN201610838804.8A CN106445701B (zh) 2016-09-21 2016-09-21 数据处理方法和装置

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US16/221,842 Continuation US11135514B2 (en) 2016-09-21 2018-12-17 Data processing method and apparatus, and storage medium for concurrently executing event characters on a game client

Publications (1)

Publication Number Publication Date
WO2018054330A1 true WO2018054330A1 (zh) 2018-03-29

Family

ID=58166723

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/102702 WO2018054330A1 (zh) 2016-09-21 2017-09-21 数据处理方法、装置和存储介质

Country Status (4)

Country Link
US (1) US11135514B2 (zh)
EP (1) EP3518105B1 (zh)
CN (1) CN106445701B (zh)
WO (1) WO2018054330A1 (zh)

Families Citing this family (19)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106445701B (zh) 2016-09-21 2018-01-09 腾讯科技（深圳）有限公司 数据处理方法和装置
CN108664842B (zh) * 2017-03-27 2020-12-18 Tcl科技集团股份有限公司 一种唇动识别模型的构建方法及系统
CN107402886B (zh) * 2017-08-09 2018-12-11 腾讯科技（深圳）有限公司 堆栈分析方法及相关装置
CN107526682B (zh) * 2017-08-16 2020-08-04 网易（杭州）网络有限公司 测试机器人的ai行为树的生成方法、装置及设备
CN109646956B (zh) * 2017-10-11 2021-08-20 腾讯科技（深圳）有限公司 信息显示方法、装置、存储介质和电子装置
TWI714078B (zh) * 2019-05-07 2020-12-21 國立高雄大學 基於深度學習之大數據分析平台排程系統及方法
CN111738443A (zh) * 2019-05-20 2020-10-02 北京京东尚科信息技术有限公司 数据处理方法、装置、系统、计算机可读存储介质
CN110170171B (zh) * 2019-06-03 2024-09-27 深圳市腾讯网域计算机网络有限公司 一种目标对象的控制方法及装置
CN110659420B (zh) * 2019-09-25 2022-05-20 广州西思数字科技有限公司 一种基于深度神经网络蒙特卡洛搜索树的个性化配餐方法
CN110826782B (zh) * 2019-10-25 2022-10-14 拉扎斯网络科技（上海）有限公司 一种数据处理的方法、装置、可读存储介质和电子设备
CN112041811B (zh) * 2019-12-12 2022-09-16 支付宝(杭州)信息技术有限公司 确定执行设备的动作选择方针
TWI725662B (zh) * 2019-12-13 2021-04-21 國立交通大學 自動化調整回合制遊戲強度之方法
CN111973985B (zh) * 2020-08-26 2024-02-09 网易（杭州）网络有限公司 基于序列的事件处理方法、装置、电子设备及存储介质
CN113791978B (zh) * 2021-09-13 2023-07-11 网易（杭州）网络有限公司 目标对照样本获取方法以及策略检测方法
CN113821513A (zh) * 2021-09-18 2021-12-21 阿里巴巴（中国）有限公司 数据处理方法、设备及存储介质
CN114021737B (zh) * 2021-11-04 2023-08-22 中国电子科技集团公司信息科学研究院 一种基于博弈的强化学习方法、系统、终端及存储介质
US20230141621A1 (en) * 2021-11-09 2023-05-11 Wonder People Co., Ltd. Method for providing battle royale game which allows players to search for sub items used for upgrading or repairing main items and game server using the same
CN114425166A (zh) * 2022-01-27 2022-05-03 北京字节跳动网络技术有限公司 数据处理方法、装置、存储介质及电子设备
CN115080241B (zh) * 2022-06-30 2024-10-15 支付宝(杭州)信息技术有限公司 数据处理方法以及装置

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020099518A1 (en) * 2001-01-23 2002-07-25 Tovinkere Vasanth R. Method and system for detecting semantic events
CN101540057A (zh) * 2009-04-24 2009-09-23 中国科学院计算技术研究所 虚拟动物驱动方法和装置
CN102136025A (zh) * 2010-12-31 2011-07-27 北京像素软件科技股份有限公司 非玩家控制角色的智能控制方法
CN106445701A (zh) * 2016-09-21 2017-02-22 腾讯科技（深圳）有限公司 数据处理方法和装置

Family Cites Families (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20010039203A1 (en) * 2000-02-23 2001-11-08 Brown Geoffrey Parker Behavior modeling in a gaming environment with contextual accuracy
US20040143852A1 (en) * 2003-01-08 2004-07-22 Meyers Philip G. Systems and methods for massively multi-player online role playing games
US20050071306A1 (en) * 2003-02-05 2005-03-31 Paul Kruszewski Method and system for on-screen animation of digital objects or characters
JP2005095403A (ja) * 2003-09-25 2005-04-14 Sega Corp ゲーム処理方法、ゲーム装置、ゲームプログラム及びその記憶媒体
US8341649B2 (en) * 2004-07-06 2012-12-25 Wontok, Inc. System and method for handling an event in a computer system
US8799857B2 (en) * 2005-04-29 2014-08-05 Microsoft Corporation XML application framework
JP4240509B2 (ja) * 2007-08-02 2009-03-18 株式会社コナミデジタルエンタテインメント ゲームシステム、端末機及びコンピュータプログラム
KR101231798B1 (ko) * 2009-04-30 2013-02-08 한국전자통신연구원 게임 난이도 조절 장치 및 방법
JP5444862B2 (ja) * 2009-06-10 2014-03-19 任天堂株式会社 ゲームプログラム、ゲーム装置、およびゲームシステム
US8959522B2 (en) * 2012-01-30 2015-02-17 International Business Machines Corporation Full exploitation of parallel processors for data processing
US9364762B2 (en) * 2013-03-14 2016-06-14 Angel Gaming, Llc Physical and environmental simulation using causality matrix
US9474974B2 (en) * 2013-12-20 2016-10-25 Gree, Inc. Method for controlling computer, recording medium and computer

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020099518A1 (en) * 2001-01-23 2002-07-25 Tovinkere Vasanth R. Method and system for detecting semantic events
CN101540057A (zh) * 2009-04-24 2009-09-23 中国科学院计算技术研究所 虚拟动物驱动方法和装置
CN102136025A (zh) * 2010-12-31 2011-07-27 北京像素软件科技股份有限公司 非玩家控制角色的智能控制方法
CN106445701A (zh) * 2016-09-21 2017-02-22 腾讯科技（深圳）有限公司 数据处理方法和装置

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP3518105A4 *

Also Published As

Publication number Publication date
EP3518105A1 (en) 2019-07-31
US20190118085A1 (en) 2019-04-25
CN106445701A (zh) 2017-02-22
EP3518105B1 (en) 2024-01-10
EP3518105A4 (en) 2019-08-14
CN106445701B (zh) 2018-01-09
US11135514B2 (en) 2021-10-05

Similar Documents

Publication Publication Date Title
WO2018054330A1 (zh) 数据处理方法、装置和存储介质
Leach Architecture in the age of artificial intelligence
US10922866B2 (en) Multi-dimensional puppet with photorealistic movement
CN110339569B (zh) 控制游戏场景中虚拟角色的方法及装置
CN109902798A (zh) 深度神经网络的训练方法和装置
CN111111220B (zh) 多人对战游戏的自对弈模型训练方法、装置和计算机设备
CN108744512A (zh) 信息提示方法和装置、存储介质及电子装置
CN109499068A (zh) 对象的控制方法和装置、存储介质、电子装置
CN108090561A (zh) 存储介质、电子装置、游戏操作的执行方法和装置
CN108269110A (zh) 基于社区问答的物品推荐方法、系统及用户设备
US11887224B2 (en) Method, apparatus, and computer program for completing painting of image, and method, apparatus, and computer program for training artificial neural network
Druce et al. Brittle AI, causal confusion, and bad mental models: challenges and successes in the XAI program
WO2023024762A1 (zh) 人工智能对象控制方法、装置、设备及存储介质
Liu et al. Graph neural network based behavior prediction to support multi-agent reinforcement learning in military training simulations
CN116977661A (zh) 一种数据处理方法、装置、设备、存储介质及程序产品
CN116943220A (zh) 一种游戏人工智能控制方法、装置、设备及存储介质
CN115487508A (zh) 游戏战队推荐模型的训练方法和相关装置
US11704980B2 (en) Method, apparatus, and computer storage medium for outputting virtual application object
CN109999497A (zh) 虚拟对象的控制方法和装置、存储介质和电子装置
WO2023246270A1 (zh) 信息处理方法、装置和存储介质及电子设备
CN116798052B (zh) 文本识别模型的训练方法和装置、存储介质及电子设备
US20240282204A1 (en) Using collected user data and cognitive information to generate a scene set in a training environment
Losh When Walls Can Talk: Animate Cities and Digital Rhetoric
de Carvalho et al. An application of genetic algorithm based on abstract data type for the problem of generation of scenarios for electronic games
CN116983639A (zh) 虚拟对象的控制方法、装置、设备及存储介质

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17852398

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2017852398

Country of ref document: EP

Effective date: 20190423