WO2021057539A1 - 一种虚拟应用对象输出方法、装置以及计算机存储介质 - Google Patents

一种虚拟应用对象输出方法、装置以及计算机存储介质 Download PDF

Info

Publication number
WO2021057539A1
WO2021057539A1 PCT/CN2020/115219 CN2020115219W WO2021057539A1 WO 2021057539 A1 WO2021057539 A1 WO 2021057539A1 CN 2020115219 W CN2020115219 W CN 2020115219W WO 2021057539 A1 WO2021057539 A1 WO 2021057539A1
Authority
WO
WIPO (PCT)
Prior art keywords
virtual application
output
state
application object
virtual
Prior art date
Application number
PCT/CN2020/115219
Other languages
English (en)
French (fr)
Inventor
蔺洁琼
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 JP2021563356A priority Critical patent/JP7408213B2/ja
Publication of WO2021057539A1 publication Critical patent/WO2021057539A1/zh
Priority to US17/460,610 priority patent/US11704980B2/en

Links

Images

Classifications

    • G—PHYSICS
    • G07—CHECKING-DEVICES
    • G07F—COIN-FREED OR LIKE APPARATUS
    • G07F17/00—Coin-freed apparatus for hiring articles; Coin-freed facilities or services
    • G07F17/32—Coin-freed apparatus for hiring articles; Coin-freed facilities or services for games, toys, sports, or amusements
    • G07F17/3202—Hardware aspects of a gaming system, e.g. components, construction, architecture thereof
    • G07F17/3204—Player-machine interfaces
    • G07F17/3211—Display means
    • G07F17/3213—Details of moving display elements, e.g. spinning reels, tumbling members
    • 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
    • A63F13/33—Interconnection arrangements between game servers and game devices; Interconnection arrangements between game devices; Interconnection arrangements between game servers using wide area network [WAN] connections
    • 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
    • A63F13/50—Controlling the output signals based on the game progress
    • A63F13/52—Controlling the output signals based on the game progress involving aspects of the displayed game scene
    • 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
    • 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
    • A63F13/60—Generating or modifying game content before or while executing the game program, e.g. authoring tools specially adapted for game development or game-integrated level editor
    • A63F13/67—Generating or modifying game content before or while executing the game program, e.g. authoring tools specially adapted for game development or game-integrated level editor adaptively or by learning from player actions, e.g. skill level adjustment or by storing successful combat sequences for re-use
    • 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
    • A63F13/80—Special adaptations for executing a specific game genre or game mode
    • G—PHYSICS
    • G07—CHECKING-DEVICES
    • G07F—COIN-FREED OR LIKE APPARATUS
    • G07F17/00—Coin-freed apparatus for hiring articles; Coin-freed facilities or services
    • G07F17/32—Coin-freed apparatus for hiring articles; Coin-freed facilities or services for games, toys, sports, or amusements
    • G07F17/3286—Type of games
    • G07F17/3293—Card games, e.g. poker, canasta, black jack
    • 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
    • A63F9/00—Games not otherwise provided for
    • A63F9/20—Dominoes or like games; Mah-Jongg games
    • A63F2009/205—Mah-jongg games

Definitions

  • This application relates to the field of Artificial Intelligence (AI), and specifically relates to virtual application object processing technology.
  • AI Artificial Intelligence
  • Online games usually use the Internet as a transmission medium, game operator servers and user computers as processing terminals, and game client software as information interaction windows. They include sustainable entertainment, leisure, communication, and virtual achievements. Sexual individual multiplayer online games.
  • the embodiments of the present application provide a virtual application object output method, device, and computer storage medium, which can improve the accuracy of target virtual application object output.
  • the embodiment of the present application provides a method for outputting a virtual application object, which is executed by a network device, and includes:
  • a virtual application object state plane is constructed, wherein the virtual application object state plane includes an area corresponding to each virtual application object, and the area includes its corresponding virtual application object.
  • the virtual application object state plane determine the output probabilities corresponding to each of the multiple virtual application objects to be output
  • a target virtual application object is determined from the multiple virtual application objects to be output for output.
  • an embodiment of the present application also provides a virtual application object output device, including:
  • An obtaining module configured to obtain current state information of multiple virtual application objects in a virtual application, where the current state information is used to indicate that the virtual application object is in a known state or an unknown state;
  • the construction module is used to construct a virtual application object state plane according to the current state information of the multiple virtual application objects, wherein the virtual application object state plane includes an area corresponding to each virtual application object, and the area is preset Arranging according to an arrangement rule, and the area includes the current state information of the corresponding virtual application object;
  • a probability determination module configured to determine the output probabilities corresponding to each of the multiple virtual application objects to be output according to the virtual application object state plane;
  • the output module is configured to determine the target virtual application object from the multiple virtual application objects to be output according to the respective output probabilities of the multiple virtual application objects to be output for output.
  • an embodiment of the present application also provides a storage medium that stores a plurality of instructions, and the instructions are suitable for loading by a processor to execute any of the virtual application object output methods provided in the embodiments of the present application. A step of.
  • embodiments of the present application also provide a computer program product, including instructions, which when run on a computer, cause the computer to execute the steps in any of the methods for outputting virtual application objects provided in the embodiments.
  • the embodiments of the present application can obtain current state information of multiple virtual application objects in a virtual application.
  • the current state information is used to indicate that the virtual application object is in a known state or an unknown state.
  • a virtual The application object state plane where the virtual application object state plane includes the area corresponding to each virtual application object, and the area includes the current state information of the corresponding virtual application object.
  • multiple virtual applications to be output are determined According to the output probabilities corresponding to each object, the target virtual application object is determined from the multiple virtual application objects to be output for output according to the output probability corresponding to each of the multiple virtual application objects to be output.
  • This solution expresses the current state information of the virtual application object in the virtual application in the state plane, so that the current state information of the virtual application object can be concisely and accurately expressed in the state plane, so as to conveniently and accurately determine the virtual application to be output
  • the output probability corresponding to the object helps to improve the accuracy of the output of the target virtual application object.
  • FIG. 1 is a schematic diagram of a scene of a virtual application object output system provided by an embodiment of the present application
  • FIG. 2 is a first flowchart of a method for outputting a virtual application object provided by an embodiment of the present application
  • FIG. 3 is a second flowchart of a method for outputting a virtual application object provided by an embodiment of the present application
  • FIG. 4 is a third flowchart of a method for outputting virtual application objects provided by an embodiment of the present application.
  • FIG. 5 is a fourth flowchart of a method for outputting a virtual application object provided by an embodiment of the present application
  • FIG. 6 is a fifth flowchart of a method for outputting a virtual application object provided by an embodiment of the present application.
  • FIG. 7 is a schematic diagram of the first interface of the network mahjong game application provided by an embodiment of the present application.
  • FIG. 8 is a schematic diagram of a second interface of a network mahjong game application provided by an embodiment of the present application.
  • FIG. 9 is a schematic diagram of a training process of a probability acquisition network provided by an embodiment of the present application.
  • FIG. 10 is a schematic diagram of a process for obtaining network output target virtual application objects with application probability according to an embodiment of the present application
  • FIG. 11 is a first schematic diagram of a first type of virtual application object state plane provided by an embodiment of the present application.
  • Figure 12 is a second schematic diagram of the first type of virtual application object state plane provided by an embodiment of the present application.
  • FIG. 13 is a third schematic diagram of the first type of virtual application object state plane provided by an embodiment of the present application.
  • FIG. 14 is a fourth schematic diagram of the first type of virtual application object state plane provided by an embodiment of the present application.
  • 15 is a first schematic diagram of a second type of virtual application object state plane provided by an embodiment of the present application.
  • 16 is a second schematic diagram of the second type of virtual application object state plane provided by an embodiment of the present application.
  • FIG. 17 is a third schematic diagram of the second type of virtual application object state plane provided by an embodiment of the present application.
  • FIG. 18 is a first schematic diagram of a third state plane of a virtual application object provided by an embodiment of the present application.
  • FIG. 19 is a second schematic diagram of a third state plane of a virtual application object provided by an embodiment of the present application.
  • FIG. 20 is a third schematic diagram of a third state plane of a virtual application object provided by an embodiment of the present application.
  • FIG. 21a is a schematic diagram of a fourth state plane of a virtual application object provided by an embodiment of the present application.
  • Figure 21b is a schematic diagram of a fifth virtual application object state plane provided by an embodiment of the present application.
  • FIG. 22 is a schematic structural diagram of a virtual application object output device provided by an embodiment of the present application.
  • FIG. 23 is a schematic diagram of the structure of a network device provided by an embodiment of the present application.
  • the embodiments of the present application provide a method, device, and computer storage medium for outputting a virtual application object.
  • the virtual application object output device can be integrated in a network device, the network device can be a terminal, a server, etc.; wherein, the terminal can be a mobile phone, a tablet computer, a notebook computer, a personal computer (PC, Personal Computer), a micro processor Boxes and other equipment; the server can be an application server or a web server. During specific deployment, it can be an independent server, a cluster server, or a cloud server.
  • Figure 1 is a schematic diagram of an application scenario of a virtual application object output method provided by an embodiment of the application.
  • the network device can acquire multiple virtual application objects in a virtual application
  • the current state information of the virtual application object is used to indicate that the virtual application object is in a known state or an unknown state.
  • a virtual application object state plane is constructed, where the virtual application object state plane includes each A region corresponding to a virtual application object, and the region includes the current state information of the corresponding virtual application object.
  • the output probability of each virtual application object to be output is determined, and the output probability is determined according to the multiple virtual application objects to be output.
  • the output probability corresponding to each object is determined from a plurality of virtual application objects to be output, and the target virtual application object is determined for output.
  • An embodiment of the present invention provides a method for outputting a virtual application object.
  • the method is executed by a server as an example for introduction.
  • the specific process of the method for outputting a virtual application object may be as follows:
  • the virtual application may be application software installed on the terminal, which can meet the application requirements of users in different fields and different problems, and provide users with a rich experience.
  • the virtual application in the embodiment of the present application may be a game application.
  • the game application may be a software product obtained by combining various programs and animation effects. Through the game application, the user’s brain, eyes, hands and other organs can be exercised. , Improve the user’s logic, agility, etc.
  • the virtual application in the embodiment of the present application may also be a card game application that needs to provide users with recommended operations, such as an online mahjong game application, an online poker game application, and so on.
  • the virtual application object may be an application-related virtual object in the virtual application.
  • the virtual application when the virtual application is a game application, the virtual application object may be a game object.
  • a virtual application object may be A virtual mahjong card; in online card game applications, a virtual application object can be a virtual card, and so on.
  • the current state information may be state information indicating the current state of the virtual application object in the virtual application.
  • the current state information may indicate the current state information of the virtual mahjong tiles, such as the current state information. It is used to indicate whether the virtual mahjong tiles are currently in a known state or an unknown state.
  • the virtual application is an online mahjong game application
  • the virtual mahjong tiles known to the current player the virtual mahjong tiles that have been output by each player, and the virtual mahjong tiles displayed by other players
  • the corresponding current state information should indicate that the virtual mahjong tiles are in a known state
  • their corresponding current state information should indicate that the virtual mahjong tiles are in an unknown state.
  • the definition method of the current state information of the virtual application object can also be adjusted according to the actual situation. For example, from the perspective of the current player, the virtual mahjong tiles known by the current player, the virtual mahjong tiles that have been output by each player, The virtual mahjong tiles displayed by other players and the virtual mahjong tiles whose specific status cannot be known are respectively recognized as different types of current status information, and so on.
  • the embodiment of the present application does not limit the definition method of the current state information, and the definition method of the current state information can be appropriately adjusted according to different virtual applications, different virtual application objects, and different game rules in virtual applications.
  • the flexibility of the output method of the virtual application object can be improved, so that it can adapt to more types of virtual applications.
  • the current status information can also be expressed in the form of identifiers, and there can be many ways to represent the current status information, for example, different colors can be used to indicate different types of current status information; or different patterns, characters, etc. can be used to indicate Different kinds of current status information, etc.
  • the current status information can be represented by the binarization flag "0" or "1". When the current status information is "0", it can indicate that the corresponding virtual mahjong tile is in an unknown state; when the current status information is "1" , Can indicate that the corresponding virtual mahjong tiles are in a known state, and so on.
  • the current state information corresponding to each virtual mahjong tile can be determined.
  • virtual mahjong tiles with known states such as virtual mahjong tiles known by the current player, virtual mahjong tiles already output by each player, and virtual mahjong tiles displayed by other players can be recognized as known.
  • State and record the current state information of the virtual mahjong tiles in the known state as "1"; while the remaining virtual mahjong tiles that cannot know the specific state can be identified as an unknown state, and record the current state information of the virtual mahjong tiles in the unknown state as "1" 0".
  • virtual mahjong tiles known by the current player virtual mahjong tiles already output by each player, virtual mahjong tiles displayed by other players, and other virtual mahjong tiles whose specific status cannot be known It is marked by different colors or different characters, and different colors or different characters are used to indicate different types of current status information, and so on.
  • Defining the current state information through a variety of methods can improve the flexibility of the virtual application object output method in the embodiments of the present application, so that it can adapt to more types of virtual applications, thereby improving the accuracy of virtual application object output.
  • the virtual application object state plane can be a plane that represents the current state of all virtual application objects in the virtual application, through which the current state of all virtual application objects can be known, and calculations can be made based on this plane to determine the output target Virtual application object.
  • the virtual application object state plane may include areas corresponding to each virtual application object. These areas are arranged according to a preset arrangement rule, and each area is used to record the virtual application object corresponding to the area. Current status information.
  • the cards in some card games will include different suits, different card types, and combinations of suits and card types.
  • the mahjong tiles in Mahjong include ten thousand, strips, tubes, wind, arrows, etc.
  • Different suits also include different card face values such as one, two, and three.
  • the combination of suit and card face value will form 34 different mahjong tiles. Therefore, the current state information of the mahjong tiles can be expressed on the plane in a binary form to facilitate subsequent network model learning.
  • a virtual application object can be constructed according to the current state information of each virtual mahjong tile State plane.
  • the virtual application object state plane includes multiple rectangular areas, and each rectangular area represents a virtual mahjong tile.
  • Multiple rectangular areas are based on the mahjong tile name of the virtual mahjong tiles corresponding to each area, in accordance with "ten thousand, twenty thousand, thirty thousand, four thousand, fifty thousand, sixty thousand, seventy thousand, eighty thousand, ninety thousand, one, two , Three, Four, Five, Six, Seven, Eight, Nine, One, Two, Three, Four, Five, Six, Seven, Eight, Nine, East, South, West, North, Arranged in the order of "medium, hair, white", the overall arrangement is a 4*34 rectangular array, where each row in the rectangular array includes 34 virtual mahjong tiles with different names, and each column in the rectangular array includes 4 tiles with the same name. Of virtual mahjong tiles.
  • Each rectangular area includes the current state information "0" or "1" of its corresponding virtual mahjong tile.
  • the rectangular area including “0” can indicate that the state of the virtual mahjong tile corresponding to the rectangular area is unknown, including "
  • the rectangular area of 1" can indicate that the state of the virtual mahjong tiles corresponding to the rectangular area is known.
  • the virtual application object state plane constructed by this method is a binary plane.
  • the preset arrangement rules of the rectangular area can also be adjusted, for example, you can also Set the area corresponding to the virtual mahjong tiles with the suit of "bar" or "tube” at the leftmost end of the plane; or in the area corresponding to the same suit in the plane, follow the virtual mahjong tiles from the largest to the smallest face value. Arrangement to the right, and so on.
  • the fan type can be the name of the combination of various cards with a certain value in mahjong or the name of the combination of cards.
  • the card type conditions meet the regulations, and meet or exceed the tie score standard, at this time, the player can be considered as the card.
  • multiple possible fan seed modes can be found in the virtual application object state plane.
  • the virtual mahjong tiles in the thick line envelope area in the figure can respectively form "one color”.
  • the state plane of virtual mahjong tiles can be expressed in such a way as "three step heights", “three colors and three simultaneous engravings", and "four happiness", so that the computer can recognize a variety of types according to the state plane. And then predict the output of the virtual application object.
  • the state plane of the virtual application object constructed according to the above method may have various misjudgments.
  • the area 1 in the figure can indicate the "one color and three step heights", and the area 2
  • the shape of the region is the same as that of region 1, it is not a representation of "one color and three step heights", which will cause certain difficulties in subsequent model learning.
  • the step of "constructing a virtual application object state plane according to the current state information of the multiple virtual application objects" may include:
  • a virtual application object state plane is constructed, wherein the virtual application object state plane includes a number of state sub-planes and a number of isolation regions, and the state sub-plane is related to the object type.
  • the isolation region is located between two adjacent state sub-planes.
  • the virtual mahjong tile can be divided into multiple object types.
  • the virtual mahjong tile can be divided into "10,000” and " There are five object types: bar, tube, wind, and arrow. Among them, the object type "10,000” can include “10,000”, “20,000”, “30,000”, “40,000”, and "10,000".
  • the object type "bar” can include "one piece", “two pieces”, and “three pieces” , “Four Tie”, “Five Tie”, “Six Tie”, “Seven Tie”, “Eight Tie”, and “Nine Tie” virtual mahjong tiles;
  • the object type “tube” can include “one tube”, “two tube”, Virtual Mahjong tiles with nine names: “Three Tubes”, “Four Tubes”, “Five Tubes”, “Six Tubes”, “Seven Tubes”, “Eight Tubes” and “Nine Tubes”;
  • the object type “Wind” can include Virtual mahjong tiles with four names of "East”, “South”, “West” and “North”;
  • the object type “Arrow” can include virtual mahjong tiles with three names of " ⁇ ", “Fat” and “White” .
  • the virtual application object state plane can be constructed according to the current state information of each virtual mahjong tile and the object type corresponding to each virtual mahjong tile, as shown in Figure 15.
  • the virtual application object state plane includes 5 state sub-planes and 4 isolated areas, of which the 5 state sub-planes correspond to five types of objects: "10,000", “bar", “tube”, “wind”, and "arrow”. Types of.
  • the isolation area is a zero-value column and is composed of 4 rectangular areas. The current state information represented in each rectangular area is 0, and the isolation area is located between two adjacent state sub-planes.
  • the state plane of the virtual application object is a 4*38 rectangular array, and the regions in the 10th, 20th, 30th, and 35th columns counted from left to right are all isolated regions. Counting from left to right, the status sub-planes from the 1st to 9th columns correspond to the object type "10,000", the status sub-planes from the 11th to 19th columns correspond to the object type "bar”, and the states from the 21st to 29th columns The sub-plane corresponds to the object type "tube", the state sub-planes in the 31st to 34th columns correspond to the object type "wind”, and the state sub-planes in the 36th to 38th columns correspond to the object type "arrow".
  • the region 1 in the virtual application object state plane can represent "one color and three step heights", and the shape of the region 2 is different from that of the region 1 due to the isolation effect of the isolation region, thereby reducing the number of types. The possibility of misjudgment.
  • the state plane of the virtual application object including the isolation area as shown in FIG. 15 can also be obtained by transforming the state plane that does not include the isolation area as shown in FIG. 12.
  • the step of "constructing a virtual application object state plane according to the current state information and the object type" may include:
  • An isolation area is inserted between two adjacent state sub-planes to obtain a virtual application object state plane.
  • the first initial state plane as shown in Figure 12 can be constructed based on the current state information of the virtual mahjong tiles, and then the first initial state plane can be divided into "10,000", “bar”, “ Five object types, “tube”, “wind”, and “arrow” are divided into five state sub-planes, each state sub-plane corresponds to one object type, and then a column of isolation is inserted in the two adjacent state sub-planes Area, the virtual application object state plane as shown in Figure 15 is obtained.
  • the state plane of the virtual application object constructed according to the above method may still have various misjudgments.
  • area three in the figure can represent various types of "one color and three uniforms", in which, Although the shape of area 4 is the same as that of area 3, it is not a representation form of "one color and three uniforms", but a representation form of "three wind carvings".
  • area 5 has the same shape as area 3, it is not a fan. This kind of representation form of "one color and three in harmony”, but a kind of representation form of "big three yuan", this will cause certain difficulties in subsequent model learning.
  • the step of "constructing a virtual application object state plane according to the current state information and the object type" may include:
  • the target object type that needs to be isolated, and the target object type includes multiple target object subtypes
  • a virtual application object state plane is constructed, wherein the state sub-plane in the virtual application object state plane includes several isolated state regions, and The isolated state area corresponds to the target object subtype, and the isolated area in the virtual application object state plane is located between two adjacent isolated state areas.
  • the virtual application is a network mahjong game application and the virtual application object is a virtual mahjong tile
  • the type determine the two target types of "wind” and “arrow” that need to be isolated, and divide the target type "wind” into four types of targets: "east”, “south”, “west”, and “north” Object subtypes; the target object type "arrow” is divided into three target object subtypes: “medium”, "fat”, and “white”.
  • a virtual application object state plane is constructed according to the current state information, object type, and target object subtype.
  • the virtual application object state plane includes 3 state sub-planes, 7 isolated state regions, and 9 isolated regions.
  • the 3 state sub-planes correspond to "10,000", "bar”, and There are three object types in the "tube”, and the 7 isolated state areas correspond to the seven target object sub-types: "East”, “South”, “West”, “North”, “Medium”, “Fa”, and "White”.
  • the isolation area is a column of 0 values, which are located between the state sub-planes corresponding to adjacent object types, between the isolated state areas corresponding to adjacent target object sub-types, and adjacent state sub-planes and isolated state sub-planes. Between status areas.
  • the virtual application object state plane is a 4*46 rectangular array, counting from left to right, the 10th, 20th, 30th, 32nd, 34th, 36th, 38th, and 38th columns from left to right.
  • the 39th, 41st, 42nd, 44th, and 45th columns are all isolated regions.
  • the status sub-planes from the 1st to 9th columns correspond to the object type "10,000”
  • the status sub-planes from the 11th to 19th columns correspond to the object type "bar”
  • the states from the 21st to 29th columns The sub-plane corresponds to the object type "tube”
  • the isolated state area in column 31 corresponds to the target object subtype "East”
  • the isolated state area in column 33 corresponds to the target object subtype "South”
  • the isolated state in column 35 The area corresponds to the target object subtype "West”
  • the isolated status area in the 37th column corresponds to the target object subtype "North”
  • the isolated status area in the 40th column corresponds to the target object subtype "Medium”
  • the 43rd column after isolation The status area corresponds to the target object subtype "Fa”
  • the isolated status area in the 46th column corresponds to the target object subtype "White”.
  • the isolation between the state sub-planes can be performed through isolation regions of different area sizes to achieve strict differentiation.
  • the purpose of the type. Specifically, the step "constructing a virtual application object state plane based on the current state information, the object type, and the target object subtype" may include:
  • a virtual application object state plane is constructed, wherein one of the two isolated state areas in the virtual application object state plane
  • the space includes the isolated area of the area size.
  • the area size of the isolation area between the target object subtypes "East” and “South” can be determined to be 1, and the area size of the isolation area between the target object subtypes “North” and “Middle” can be determined Is 2, the area size of the isolated area between the target object subtypes " ⁇ " and “ ⁇ ” is 2, and so on.
  • the virtual application object state plane can be constructed according to the current state information, object type, target object subtype, and area size.
  • the number of columns in the isolated area between the isolated state area corresponding to "East” and the isolated state area corresponding to “South” is 1, while the isolated state area corresponding to "Middle” and the isolated state area corresponding to "Fa”
  • the number of columns of isolation regions between the corresponding isolated state regions is 2.
  • the number of columns in the isolation area is not excessively limited, as long as it is ensured that the number of columns in the isolation area is different between the isolated state areas corresponding to the target object subtypes that need to be strictly distinguished.
  • area three in the virtual application object state plane can represent "one color and three in the same order".
  • Area four has a different shape from area three due to the isolation effect of the isolation areas with different numbers of columns. It means the "three wind carvings" of different kinds.
  • the shape of area five is different from the shapes of areas three and four due to the isolation effect of the isolation area with different numbers of columns. It can express the kinds of "big three yuan", thereby reducing the error of the type. The possibility of judgment.
  • the virtual application object state plane including isolation regions with different numbers of columns can also be transformed by transforming the state plane including isolation regions with the same number of columns as shown in FIG. get.
  • the step of "constructing a virtual application object state plane according to the current state information, the object type, the target object subtype, and the area size" may include:
  • a virtual application object state plane is constructed.
  • a second initial state plane as shown in Figure 15 can be constructed according to the current state information and the object type, and then the target state corresponding to the target object type "wind” and the target object type “arrow” can be determined. Plane, and then divide the target state sub-plane corresponding to the target object type "wind” into four isolated state regions, corresponding to the four target object subtypes of "East”, “South”, “West”, and “North”; The target state sub-plane corresponding to the target object type "arrow” is divided into three isolated state regions, corresponding to the three target object subtypes of "medium”, "fat”, and “white” respectively.
  • the plane can also be rotated to make the length and width gap smaller.
  • the state plane of the virtual application object in the figure includes 5 rotated state sub-planes and multiple isolated regions. Each rotated state sub-plane corresponds to one object type, and the isolated regions will be different objects. Types, and different target object sub-types are isolated.
  • the virtual application object state plane is a 9*24 rectangular array.
  • the vertical direction represents the face value of the virtual mahjong tiles from 1 to 9, and the horizontal direction corresponds to "ten thousand", “bar”, “tube”, “wind”, and The five rotated state sub-planes of "Arrow". It should be understood that after the plane is rotated, the state plane as shown in FIG. 21b can also be obtained, that is, a 24*9 rectangular array can be obtained.
  • the state plane of the virtual application object shown in FIG. 21a and FIG. 21b may also be obtained by transforming the state plane shown in FIG. 18.
  • the step of "constructing a virtual application object state plane according to the current state information, the object type, the target object subtype, and the area size" may include:
  • An isolation region is inserted between two adjacent rotated state sub-planes to obtain a virtual application object state plane.
  • the third initial state plane as shown in Figure 20 can be constructed according to the current state information, object type, target object subtype, and area size. ",” “wind”, and “arrow” five object types, the third initial state plane is divided into 5 sub-planes to be rotated and 3 isolated regions. Then rotate each sub-plane of the state to be rotated to obtain 5 sub-planes of the rotated state, and add a 2*4 isolation area to the rotated sub-plane corresponding to the "arrow", so that the 5 sub-planes of the rotated state can be Splice into a rectangle, and then insert a column of isolation regions in the two adjacent rotated state sub-planes to obtain the virtual application object state plane.
  • FIGS. 21a and 21b there are multiple methods for constructing the virtual application object state plane as shown in FIGS. 21a and 21b, and they are not limited to the methods given above.
  • the isolation area is inserted from the width direction and the height direction respectively to reduce the possibility of confusion among multiple seed modes in the state plane, thereby reducing the possibility of misjudgment. Sex. And the whole state plane is closer to a square, which is conducive to the learning of the neural network.
  • the virtual application object state plane determine the output probabilities corresponding to each of the multiple virtual application objects to be output.
  • the virtual application object to be output may be a virtual application object that is set according to game rules and can be output.
  • the virtual application when the virtual application is a network mahjong game application and the virtual application object is a virtual mahjong tile, the virtual application object to be output may be a virtual mahjong tile that can be played by the current player.
  • the output probability of each virtual application object to be output can be determined according to the virtual application object state plane. As shown in Figure 7, the output probability of each virtual application object to be output can be determined. After the output probability corresponding to the virtual application object is to be output, it can also be displayed in the form of a table on the game interface. For example, the probability that the player north wind outputs the virtual application object to be output "south" is 62.54%, and the player north wind outputs the virtual application object to be output. The probability of the application being "medium" is 25%, and so on.
  • the output probability corresponding to each virtual application object to be output can be displayed on the game interface in the form of a list, and the virtual application object to be output with a low probability can also be omitted. And so on, to improve the flexibility of the game interface display.
  • the state plane of the virtual application object may also be calculated through a neural network.
  • the step of "determining respective output probabilities corresponding to multiple virtual application objects to be output according to the virtual application object state plane" may include:
  • the output probability corresponding to the virtual application object to be output is acquired.
  • the probability acquisition network can be a neural network
  • the neural network can be a mathematical model of an algorithm that performs distributed parallel information processing by imitating the behavior characteristics of an animal neural network.
  • the neural network relies on the complexity of the system and achieves the purpose of processing information by adjusting the interconnection between a large number of internal nodes.
  • the probability acquisition network is a CNN (Convolutional Neural Networks, convolutional neural network) model as an example for description.
  • the basic components of the convolutional neural network model include convolutional layer, pooling layer, fully connected layer, etc., among them, convolutional layer, pooling layer, etc. can form a convolution block, and multiple convolutional blocks are connected to multiple fully connected layers.
  • the connection layer forms a convolutional neural network structure.
  • the probability acquisition network may also be, for example, RNN (Recurrent Neural Network), DNN (Deep Neural Network, Deep Neural Network), random forest model, SVM (Support Vector Machine, Support vector machine) and other multi-classification model frameworks, and this embodiment is not limited to this.
  • RNN Recurrent Neural Network
  • DNN Deep Neural Network, Deep Neural Network
  • random forest model SVM (Support Vector Machine, Support vector machine) and other multi-classification model frameworks, and this embodiment is not limited to this.
  • the virtual application object state plane can be input to CNN (Convolutional Neural Networks).
  • the convolutional neural network can be called a probability acquisition network. After the calculation of the probability acquisition network, it can be calculated Obtain the output probability of the virtual application object to be output that the current player can control the output.
  • the virtual application object state plane can be regarded as a matrix of pixel values. Since the virtual application object state plane only includes two current state information of 0 and 1, therefore, the virtual application object state plane is a kind of two kinds of current state information. Value image.
  • the probability acquisition network can include several convolutional layers, and the input image can be convolved through the convolutional layers.
  • the convolution operation can learn the characteristics of the image from the input image data and preserve the spatial relationship between pixels.
  • the convolution kernel moves in the input image with a certain step size, and performs the convolution operation, and then can output the corresponding features of the image. Therefore, after the virtual application object state plane is input into the probability acquisition network, the convolutional layer in the network can be acquired through probability, and the features of the virtual application object state plane can be extracted.
  • the fully connected layer can determine the output probability corresponding to the virtual application object to be output.
  • the fully connected layer can map the learned features to the sample label space, and play a role in the network model.
  • Each node of the fully connected layer is connected to the output node of the previous layer. Among them, a node of the fully connected layer is called a neuron in the fully connected layer.
  • the number of neurons in the fully connected layer can be based on It depends on the needs of the actual application.
  • the core operation of fully connected is matrix-vector product, which is essentially a linear transformation from one feature space to another feature space. In the probabilistic acquisition network, the fully connected layer can be located in the last few layers and is used to weight the previously acquired features.
  • each neuron in the fully connected layer may also use the ReLU function.
  • the output value of the last fully connected layer is passed to an output, which can be classified by the softmax layer. After classification, the corresponding output probability of each virtual application object to be output that the current player can control the output can be calculated.
  • the probability acquisition network may also be trained.
  • the virtual application object output method may further include:
  • the preset probability acquisition network is converged to obtain the probability acquisition network.
  • the previously stored game logs can be obtained, and the obtained game logs can be sorted.
  • the game log can record the process of the game from the beginning to the end.
  • the information in the game log can be organized into multiple sample state images.
  • the target game actions can be used as real output objects.
  • the game action that the current player outputs "in” can be recognized as the target game action.
  • obtain multiple sample state images before the target game action For example, 100 sample state images before the target game action can be obtained. At this time, these 100 sample state images all correspond to the target game action, that is, a real
  • the output object can correspond to multiple sample state planes.
  • the representation form of the sample state plane may be the same as the representation form of the virtual application object state plane.
  • the preset probability acquisition network can be trained according to the obtained multiple sample state planes, and the predicted output object corresponding to the sample state plane can be obtained.
  • the preset probability The model parameters of the probability acquisition network converge.
  • the calculated matching degree between the predicted output object and the real output object meets the preset threshold, it means that the probability acquisition network at this time has been trained, and the trained probability acquisition can be obtained.
  • the output probabilities corresponding to each of the multiple virtual application objects to be output determine a target virtual application object from the multiple virtual application objects to be output for output.
  • the probability acquisition network can output the output probability corresponding to each virtual application object to be output, and then it can sample according to the acquired probability distribution to determine the target virtual application object from multiple virtual application objects to be output , And output the target virtual application object.
  • the network device to output the target virtual application object.
  • the virtual application object output method when applied to the virtual player on the computer side, it can directly help the virtual player on the computer side to output the target. Virtual application objects, thereby advancing the game process.
  • the virtual mahjong tiles corresponding to the target virtual application object in the game interface can be highlighted, for example, the virtual mahjong tiles corresponding to the target virtual application object can be highlighted.
  • the mahjong tiles flash or mark the virtual mahjong tiles with different colors.
  • the user can know that outputting this virtual mahjong tile can advance the game process, and the user can choose to output a specially marked virtual mahjong tile.
  • the user You can also choose other virtual mahjong tiles for output for other considerations, thereby enhancing the flexibility of the game.
  • the game may not be over.
  • the current state information can be updated, and the target virtual application object can be determined again.
  • the step "determine a target virtual application object for output from the plurality of virtual application objects to be output according to the respective output probabilities of the plurality of virtual application objects to be output" it may further include:
  • the virtual application is a network mahjong game application and the virtual application object is a virtual mahjong tile
  • multiple virtual application objects to be output in the virtual application have changed, that is, , The cards that have been played cannot be output again.
  • the virtual application object to be output can be updated.
  • analyze the updated virtual application objects to be output For example, it is possible to analyze the multiple virtual application objects to be output that the current player can output at this time.
  • these virtual application objects to be output cannot form the card type of a draw card, It can be considered that the virtual application object to be output does not meet the termination condition.
  • the step of constructing the virtual application object state plane based on the current state information of multiple virtual application objects can be returned to continue to determine the target virtual application object; when these virtual application objects are to be output
  • the virtual application object can form the card type of the harmony card, the virtual application object to be output can be considered to meet the termination condition.
  • the virtual application object output method can be applied to various game rules such as “xx Chenghe”, “xx to the end”, and “xx five-star” due to its accuracy and flexibility.
  • the virtual application can also be applied to a variety of game forms such as “single player”, “two-player game”, “three-player game”, “four-player game”, etc. Users can team up with other users to play games to enhance the fun of the game Sex.
  • the embodiment of the present application can obtain current state information of multiple virtual application objects in a virtual application.
  • the current state information is used to indicate that the virtual application object is in a known state or an unknown state, according to the current state of the multiple virtual application objects.
  • Information construct a virtual application object state plane, where the virtual application object state plane includes the area corresponding to each virtual application object, and the area includes the current state information of its corresponding virtual application object.
  • multiple The output probability corresponding to each virtual application object to be output is determined from the multiple virtual application objects to be output according to the output probability corresponding to each of the multiple virtual application objects to be output, and the target virtual application object is determined for output.
  • This solution expresses the current state information of each virtual application object in the virtual application in a state plane, so that the current state information of each virtual application object can be concisely and accurately expressed in the state plane, thereby facilitating the recognition and learning of the neural network To determine the output probability corresponding to the virtual object to be output.
  • This solution can also make the whole state plane closer to a square by rotating the sub-plane of the state to be rotated, thereby facilitating the learning of the neural network.
  • the solution strictly distinguishes the virtual application objects that need to be distinguished by inserting the isolation area in the state plane, which reduces the possibility of misjudgment and improves the accuracy of the output of the target virtual application object. .
  • the specific process of the virtual application object output method of the embodiment of the present application may be as follows:
  • a network device obtains current state information of multiple virtual mahjong tiles in a network mahjong game application.
  • the network device can determine whether each virtual mahjong tile is in a known state or an unknown state according to the current hand in the network mahjong game application, and can define the virtual player's own known virtual state.
  • Mahjong tiles, virtual mahjong tiles that have been output by each player, and virtual mahjong tiles displayed by other players and other virtual mahjong tiles with known states are in a known state.
  • the current state information corresponding to the virtual mahjong tiles can be "1"; the virtual mahjong tiles in other positions can be defined as unknown states.
  • the virtual mahjong tiles correspond to The current status information of can be "0".
  • the network device constructs a virtual application object state plane according to the current state information of the multiple virtual mahjong tiles.
  • the network device can construct a virtual application object state plane as shown in FIG. 12 according to the current state information corresponding to each virtual mahjong tile.
  • the virtual application object state plane includes multiple rectangular areas, and each rectangular area represents a virtual mahjong tile.
  • the multiple rectangular areas are in accordance with "ten thousand, twenty thousand, thirty thousand, four thousand, fifty thousand, sixty thousand, seventy thousand, eighty thousand, ninety thousand, one, two, Three, four, five, six, seven, eight, nine, one, two, three, four, five, six, seven, eight, nine, east, south, west, north, middle Arranged in the order of ", hair, white", the overall arrangement is a 4*34 rectangular array, where each row in the rectangular array includes 34 virtual mahjong tiles with different names, and each column in the rectangular array includes 4 tiles with the same name. Virtual Mahjong tiles. Each rectangular area includes the current state information "0" or "1" of the virtual mahjong tiles corresponding to the rectangular area.
  • the network device inputs the virtual application object state plane into the probability acquisition network, and acquires the output probabilities corresponding to each of the multiple virtual mahjong tiles to be output.
  • the virtual application object state plane can be input to the CNN network.
  • the CNN network can be a probability acquisition network, and the current player can be obtained.
  • the output probability corresponding to each virtual mahjong tile to be output is shown in Figure 7.
  • the probability that the current player north wind outputs the virtual mahjong tile "south” is 62.54%, and the probability that the current player north wind outputs the virtual mahjong tile "middle”
  • the probability that the current player North Wind will output the virtual mahjong tile "North” is 7.76%.
  • the probability that the current player North Wind will output the virtual mahjong tile "West” is 4.34%.
  • the current player North Wind will output the virtual mahjong tile "East”.
  • the probability is 0.21%, the probability that the current player North Wind will output the virtual mahjong tile "Yaoji” is 0.13%, and so on.
  • the network device determines a target virtual mahjong tile from the multiple virtual mahjong tiles to be output according to the respective output probabilities of the multiple virtual mahjong tiles to be output for output.
  • the probability distribution can be sampled to determine the target virtual mahjong tile from the multiple virtual mahjong tiles to be output, and the target virtual mahjong tile can be determined Perform output.
  • the network device updates the virtual mahjong tiles to be output.
  • the network device outputs the target virtual mahjong tiles
  • the virtual mahjong tiles to be output are changed.
  • the virtual mahjong tiles to be output can be updated to analyze the updated virtual mahjong tiles to be output.
  • the network device updates the current state information of the virtual mahjong tiles.
  • the network device can check the current status of each virtual mahjong tile in the current game according to the current game status. Information is updated.
  • the network device returns and executes the step of constructing a virtual application object state plane according to the current state information of the multiple virtual application objects.
  • each virtual mahjong tile when the current state information of each virtual mahjong tile is updated, it can return to the step of constructing a virtual application object state plane based on the current state information of multiple virtual application objects, and continue to identify the target virtual application object , Until the current player can control the virtual mahjong tiles to be output to form a sum card type, at this time, the game flow can be terminated.
  • the embodiment of the present application can obtain the current state information of multiple virtual mahjong tiles in a network mahjong game application through a network device, and construct a virtual application object state plane according to the current state information of the multiple virtual mahjong tiles, and combine the virtual application objects
  • the state plane is input to the probability acquisition network to obtain the output probabilities corresponding to the multiple virtual mahjong tiles to be output, and determine the target virtual mahjong from the multiple virtual mahjong tiles to be output according to the corresponding output probabilities of the multiple virtual mahjong tiles to be output.
  • the tiles are output and the virtual mahjong tiles to be output are updated.
  • the current state information of the virtual mahjong tiles is updated, and the execution is returned.
  • the virtual application object state is constructed Plane steps. This solution expresses the current state information of the virtual application object in the virtual application in the state plane, so that the current state information of the virtual application object can be represented in the state plane concisely and accurately, thereby facilitating the recognition and learning of the neural network.
  • the output probability corresponding to the virtual application object to be output is determined, thereby improving the accuracy of the output of the target virtual application object.
  • the specific process of the virtual application object output method of the embodiment of the present application may be as follows:
  • the network device obtains current state information of multiple virtual mahjong tiles in a network mahjong game application, and object types of the multiple virtual mahjong tiles.
  • the network device can obtain the current status information "0" or "1" of each virtual mahjong tile in the network mahjong game application, and divide the virtual mahjong tiles into “0” or “1” according to the name of each virtual mahjong tile.
  • the network device constructs a virtual application object state plane according to the current state information of the multiple virtual mahjong tiles and the object type.
  • the network device can construct a virtual application object state plane as shown in FIG. 15 according to the current state information corresponding to each virtual mahjong tile and the object type.
  • the state plane of the virtual application object is a 4*38 rectangular array, and the areas in the 10th column, the 20th column, the 30th column, and the 35th column from left to right are all isolated regions.
  • the status sub-planes from the 1st to 9th columns correspond to the object type "10,000”
  • the status sub-planes from the 11th to 19th columns correspond to the object type "bar”
  • the states from the 21st to 29th columns The sub-plane corresponds to the object type "tube”
  • the state sub-planes in the 31st to 34th columns correspond to the object type "wind”
  • the state sub-planes in the 36th to 38th columns correspond to the object type "arrow”.
  • the network device inputs the virtual application object state plane into the probability acquisition network, and acquires the output probabilities corresponding to each of the multiple virtual mahjong tiles to be output.
  • the network device determines a target virtual mahjong tile from the multiple virtual mahjong tiles to be output according to the respective output probabilities of the multiple virtual mahjong tiles to be output for output.
  • the network device updates the virtual mahjong tiles to be output.
  • the network device updates the current state information of the virtual mahjong tiles.
  • the network device returns to execute the step of constructing a virtual application object state plane according to the current state information.
  • the embodiment of the application can obtain the current state information of multiple virtual mahjong tiles in the network mahjong game application and the object types of the multiple virtual mahjong tiles through a network device, and construct a virtual application based on the current state information and object types Object state plane, input the virtual application object state plane into the probability acquisition network to obtain the output probability of the virtual mahjong tiles to be output, and determine the target virtual mahjong from multiple virtual mahjong tiles to be output according to the output probability of the virtual mahjong tiles to be output The tiles are output, and the virtual mahjong tiles to be output are updated.
  • the current state information of the virtual mahjong tiles is updated, and the step of constructing a virtual application object state plane based on the current state information is returned.
  • This solution expresses the current state information of the virtual application object in the virtual application in the state plane, so that the current state information can be concisely and accurately expressed in the state plane, thereby facilitating the recognition and learning of the neural network, and determining the virtual output to be output.
  • the output probability corresponding to the application object further improves the accuracy of the output of the target virtual application object.
  • the specific process of the virtual application object output method of the embodiment of the present application may be as follows:
  • the network device obtains the current status information of multiple virtual mahjong tiles in the network mahjong game application, the object types of multiple virtual mahjong tiles, the target object subtype that needs to be isolated, and the isolation area between the two target object subtypes Area size.
  • the network device can determine the five object types “Wan”, “Bar”, “Cylinder”, “Wind”, and “Arrow”, among which the target object types “Wind” and “Arrow” need to be isolated. Then the object type "wind” is divided into four target object subtypes: “east”, “south”, “west”, and “north”, and the object type “arrow” is divided into “middle”, “fat”, and “ “White” three target object sub-types.
  • the network device constructs a virtual application object state plane according to the current state information, object type, target object subtype, and area size.
  • the network device can construct a virtual application object state plane as shown in FIG. 18 according to the current state information, object type, target object subtype, and area size corresponding to each virtual mahjong tile.
  • the state plane of the virtual application object is a 4*46 rectangular array, counting from left to right, the 10th, 20th, 30th, 32nd, 34th, 36th, 38th, 39th columns. , The 41st, 42nd, 44th, and 45th columns are all isolated regions.
  • the status sub-planes from the 1st to 9th columns correspond to the object type "10,000”
  • the status sub-planes from the 11th to 19th columns correspond to the object type "bar”
  • the states from the 21st to 29th columns The sub-plane corresponds to the object type "tube”
  • the isolated state area in column 31 corresponds to the target object subtype "East”
  • the isolated state area in column 33 corresponds to the target object subtype "South”
  • the isolated state in column 35 The area corresponds to the target object subtype "West”
  • the isolated status area in the 37th column corresponds to the target object subtype "North”
  • the isolated status area in the 40th column corresponds to the target object subtype "Medium”
  • the 43rd column after isolation The status area corresponds to the target object subtype "Fa”
  • the isolated status area in the 46th column corresponds to the target object subtype "White”.
  • the network device inputs the virtual application object state plane into the probability acquisition network, and acquires the output probability of the virtual mahjong tiles to be output.
  • the network device determines the target virtual mahjong tiles from the multiple virtual mahjong tiles to be output for output.
  • the network device updates the virtual mahjong tiles to be output.
  • the network device updates the current state information of the virtual mahjong tiles.
  • the network device returns to execute the step of constructing a virtual application object state plane according to the current state information.
  • the embodiment of the present application can obtain the current state information of multiple virtual mahjong tiles in the network mahjong game application, the object types of the multiple virtual mahjong tiles, the target object subtypes that need to be isolated, and two targets through a network device.
  • the area size of the isolated area between the object subtypes construct a virtual application object state plane, input the virtual application object state plane into the probability acquisition network, and obtain Output probability of virtual mahjong tiles, according to the output probability of the virtual mahjong tiles to be output, determine the target virtual mahjong tiles from multiple virtual mahjong tiles to be output for output, update the virtual mahjong tiles to be output, and when the virtual mahjong tiles to be output are not satisfied
  • the current state information of the virtual mahjong tiles is updated, and the step of constructing the state plane of the virtual application object according to the current state information is returned.
  • This solution expresses the current state information of the virtual application object in the virtual application in the state plane, so that the current state information can be concisely and accurately expressed in the state plane, thereby facilitating the recognition and learning of the neural network, and determining the virtual output to be output.
  • the solution strictly distinguishes the virtual application objects that need to be distinguished by inserting the isolation area in the state plane, reduces the possibility of misjudgment, and improves the accuracy of the output of the target virtual application object.
  • the specific process of the virtual application object output method of the embodiment of the present application may be as follows:
  • the network device obtains the current state information of multiple virtual mahjong tiles in the network mahjong game application, the object types of multiple virtual mahjong tiles, the target object subtype that needs to be isolated, and the isolation area between the two target object subtypes Area size.
  • the network device constructs a virtual application object state plane according to the current state information, object type, target object subtype, and area size.
  • the network device can construct a virtual application object state plane as shown in FIG. 21a according to the current state information, object type, target object subtype, and area size corresponding to each virtual mahjong tile.
  • the virtual application object state plane includes 5 rotated state sub-planes and multiple isolation regions. Each rotated state sub-plane corresponds to one object type. The isolation regions will have different object types and different target object subtypes. Isolate.
  • the virtual application object state plane is a 9*24 rectangular array.
  • the vertical direction represents the face value of the virtual mahjong tiles from 1 to 9, and the horizontal direction corresponds to "ten thousand", "bar", "tube”, “wind”, and The five rotated state sub-planes of "Arrow".
  • the network device inputs the virtual application object state plane into the probability acquisition network to acquire the output probability of the virtual mahjong tiles to be output.
  • the network device determines the target virtual mahjong tiles from the multiple virtual mahjong tiles to be output according to the output probability of the virtual mahjong tiles to be output for output.
  • the network device updates the virtual mahjong tiles to be output.
  • the network device updates the current state information of the virtual mahjong tiles.
  • the network device returns to execute the step of constructing a virtual application object state plane according to the current state information.
  • the embodiment of the present application can obtain the current state information of multiple virtual mahjong tiles in the network mahjong game application, the object types of the multiple virtual mahjong tiles, the target object subtypes that need to be isolated, and two targets through a network device.
  • the area size of the isolated area between the object subtypes construct a virtual application object state plane, input the virtual application object state plane into the probability acquisition network, and obtain Output probability of virtual mahjong tiles, according to the output probability of the virtual mahjong tiles to be output, determine the target virtual mahjong tiles from multiple virtual mahjong tiles to be output for output, update the virtual mahjong tiles to be output, and when the virtual mahjong tiles to be output are not satisfied
  • the current state information of the virtual mahjong tiles is updated, and the step of constructing the state plane of the virtual application object according to the current state information is returned.
  • This solution expresses the current state information of the virtual application object in the virtual application in the state plane, so that the current state information can be concisely and accurately expressed in the state plane, thereby facilitating the recognition and learning of the neural network, and determining the virtual output to be output.
  • the output probability corresponding to the application object can also make the whole state plane closer to a square by rotating the sub-plane of the state to be rotated, thereby facilitating the learning of the neural network.
  • the solution strictly distinguishes the virtual application objects that need to be distinguished by inserting the isolation area in the state plane, which reduces the possibility of misjudgment and improves the accuracy of the output of the target virtual application object. .
  • embodiments of the present application may also provide a virtual application object output device.
  • the virtual application object output device may be specifically integrated in a network device.
  • the network device may include a server, a terminal, etc., wherein the terminal It can include: mobile phones, tablet computers, notebook computers or personal computers (PC, Personal Computer), etc.
  • the virtual application object output device may include an acquisition module 221, a construction module 222, a probability determination module 223, and an output module 224, as follows:
  • the obtaining module 221 is configured to obtain current state information of multiple virtual application objects in a virtual application, where the current state information is used to indicate that the virtual application object is in a known state or an unknown state;
  • the construction module 222 is configured to construct a virtual application object state plane according to the current state information of the multiple virtual application objects, where the virtual application object state plane includes an area corresponding to each virtual application object, and the area includes The current state information of the corresponding virtual application object;
  • the probability determination module 223 is configured to determine the output probabilities corresponding to each of the multiple virtual application objects to be output according to the virtual application object state plane;
  • the output module 224 is configured to determine a target virtual application object to output from the multiple virtual application objects to be output according to respective output probabilities of the multiple virtual application objects to be output.
  • the construction module 222 may include a first acquisition sub-module 2221 and a first construction sub-module 2222, as follows:
  • the first obtaining submodule 2221 is configured to obtain the object types of multiple virtual application objects
  • the first construction sub-module 2222 is configured to construct a virtual application object state plane according to the current state information and the object type, wherein the virtual application object state plane includes several state sub-planes and several isolation regions, so The state sub-plane corresponds to the object type, and the isolation area is located between two adjacent state sub-planes.
  • the first construction submodule 2222 may be specifically used for:
  • An isolation area is inserted between two adjacent state sub-planes to obtain a virtual application object state plane.
  • the first construction sub-module 2222 may include a first determination sub-module 22221 and a second construction sub-module 22222, as follows:
  • the first determining submodule 22221 is configured to determine the target object type that needs to be isolated from among several object types, and the target object type includes multiple target object subtypes;
  • the second construction sub-module 22222 is configured to construct a virtual application object state plane according to the current state information, the object type, and the target object subtype, wherein the state sub-plane in the virtual application object state plane It includes several isolated state areas, the isolated state area corresponds to the target object subtype, and the isolated area in the virtual application object state plane is also located between two adjacent isolated state areas.
  • the second construction sub-module 22222 may include a second determination sub-module 222221 and a third construction sub-module 222222, as follows:
  • the second determining submodule 222221 is used to determine the area size of the isolation area between the two target object subtypes that need to be isolated;
  • the third construction sub-module 222222 is configured to construct a virtual application object state plane according to the current state information, the object type, the target object subtype, and the area size, wherein the virtual application object state plane Between the two isolated state regions in, an isolation region of the region size is included.
  • the third construction submodule 222222 may be specifically used for:
  • a virtual application object state plane is constructed.
  • the third construction submodule 222222 may be specifically used for:
  • An isolation region is inserted between two adjacent rotated state sub-planes to obtain a virtual application object state plane.
  • the virtual application object output device may further include a first update module 2251, a second update module 2252, a return module 2253, and a termination module 2254, as follows:
  • the first update module 2251 is used to update the virtual application object to be output
  • the second update module 2252 is configured to update the current state information of the virtual application object when the virtual application object to be output does not meet the termination condition;
  • the return module 2253 is used to return to execute the step of constructing a virtual application object state plane according to the current state information of the multiple virtual application objects;
  • the termination module 2254 is configured to terminate the output when the virtual application object to be output meets the termination condition.
  • the probability acquisition module 223 may be specifically used for:
  • the output probability corresponding to the virtual application object to be output is acquired.
  • the virtual application object output device may further include a second acquisition sub-module 2261, a training sub-module 2262, and a convergence sub-module 2263, as follows:
  • the second acquisition sub-module 2261 is used to acquire several sample state planes and the real output object corresponding to each sample state plane;
  • the training sub-module 2262 is configured to train the preset probability acquisition network according to the sample state plane to obtain the predicted output object corresponding to the sample state plane;
  • the convergence sub-module 2263 is configured to converge the preset probability acquisition network according to the real output object corresponding to the sample state plane and the predicted output object to obtain the probability acquisition network.
  • each of the above units can be implemented as an independent entity, or can be combined arbitrarily, and implemented as the same or several entities.
  • each of the above units please refer to the previous method embodiments, which will not be repeated here.
  • the virtual application object output device of this embodiment obtains the current state information of multiple virtual application objects in the virtual application through the obtaining module 221.
  • the current state information is used to indicate that the virtual application object is in a known state or an unknown state
  • the construction module 222 constructs a virtual application object state plane according to the current state information of multiple virtual application objects, where the virtual application object state plane includes the area corresponding to each virtual application object, and the area includes the current state of its corresponding virtual application object.
  • the probability determination module 223 determines the output probabilities corresponding to the multiple virtual application objects to be output.
  • the output module 224 determines the output probabilities corresponding to the multiple virtual application objects to be output.
  • the target virtual application object is determined for output.
  • This solution expresses the current state information of the virtual application object in the virtual application in the state plane, so that the current state information can be concisely and accurately expressed in the state plane, thereby facilitating the recognition and learning of the neural network, and determining multiple waits. Output the corresponding output probability of each virtual application object.
  • This solution can also make the whole state plane closer to a square by rotating the sub-plane of the state to be rotated, thereby facilitating the learning of the neural network.
  • the solution strictly distinguishes the virtual application objects that need to be distinguished by inserting the isolation area in the state plane, which reduces the possibility of misjudgment and improves the accuracy of the output of the target virtual application object. .
  • the embodiments of the present application also provide a network device, which can integrate any of the virtual application object output devices provided in the embodiments of the present application.
  • FIG. 23 shows a schematic structural diagram of a network device involved in an embodiment of the present application, specifically:
  • the network device may include one or more processing core processors 231, one or more computer-readable storage media storage 232, power supply 233, input unit 234 and other components.
  • processing core processors 231 one or more computer-readable storage media storage 232
  • power supply 233 input unit 234
  • other components other components that can be included in the network device.
  • FIG. 23 does not constitute a limitation on the network device, and may include more or fewer components than shown in the figure, or combine certain components, or different component arrangements. among them:
  • the processor 231 is the control center of the network device. It uses various interfaces and lines to connect the various parts of the entire network device, runs or executes the software programs and/or modules stored in the memory 232, and calls the data stored in the memory 232. Data, perform various functions of network equipment and process data, so as to monitor the network equipment as a whole.
  • the processor 231 may include one or more processing cores; preferably, the processor 231 may integrate an application processor and a modem processor, where the application processor mainly processes the operating system, user interface, application programs, etc. , The modem processor mainly deals with wireless communication. It can be understood that the foregoing modem processor may not be integrated into the processor 231.
  • the memory 232 may be used to store software programs and modules.
  • the processor 231 executes various functional applications and data processing by running the software programs and modules stored in the memory 232.
  • the memory 232 may mainly include a storage program area and a storage data area.
  • the storage program area may store an operating system, an application program required by at least one function (such as a sound playback function, an image playback function, etc.), etc.; Data created by the use of network equipment, etc.
  • the memory 232 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one magnetic disk storage device, a flash memory device, or other volatile solid-state storage devices.
  • the memory 232 may further include a memory controller to provide the processor 231 to access the memory 232.
  • the network device also includes a power supply 233 for supplying power to various components.
  • the power supply 233 may be logically connected to the processor 231 through a power management system, so that functions such as charging, discharging, and power consumption management can be managed through the power management system.
  • the power supply 233 may also include any components such as one or more DC or AC power supplies, a recharging system, a power failure detection circuit, a power converter or inverter, and a power status indicator.
  • the network device may further include an input unit 234, which can be used to receive inputted digital or character information and generate keyboard, mouse, joystick, optical or trackball signal input related to user settings and function control.
  • an input unit 234 which can be used to receive inputted digital or character information and generate keyboard, mouse, joystick, optical or trackball signal input related to user settings and function control.
  • the network device may also include a display unit, etc., which will not be repeated here.
  • the processor 231 in the network device will load the executable file corresponding to the process of one or more application programs into the memory 232 according to the following instructions, and the processor 231 will run and store the executable file in the memory 232.
  • the application programs in the memory 232 are as follows:
  • the current state information is used to indicate that the virtual application object is in a known state or an unknown state.
  • a virtual application object state plane is constructed, Among them, the virtual application object state plane includes the area corresponding to each virtual application object, and the area includes the current state information of its corresponding virtual application object.
  • the corresponding output of the multiple virtual application objects to be output is determined. Probability, according to the output probabilities corresponding to each of the multiple virtual application objects to be output, the target virtual application object is determined from the multiple virtual application objects to be output for output.
  • the network device of the embodiment of the present application can obtain current state information of multiple virtual application objects in a virtual application.
  • the current state information is used to indicate that the virtual application object is in a known state or an unknown state, according to the multiple virtual application objects.
  • Construct a virtual application object state plane where the virtual application object state plane includes the area corresponding to each virtual application object, and the area includes the current state information of its corresponding virtual application object.
  • the output probabilities corresponding to the multiple virtual application objects to be output are determined, and the target virtual application objects are determined for output from the multiple virtual application objects to be output according to the output probabilities corresponding to the multiple virtual application objects to be output.
  • This solution expresses the current state information of the virtual application object in the virtual application in the state plane, so that the current state information can be concisely and accurately expressed in the state plane, thereby facilitating the recognition and learning of the neural network to determine the output to be output
  • the output probability corresponding to the virtual application object can also make the whole state plane closer to a square by rotating the sub-plane of the state to be rotated, thereby facilitating the learning of the neural network.
  • the solution strictly distinguishes the virtual application objects that need to be distinguished by inserting the isolation area in the state plane, which reduces the possibility of misjudgment and improves the accuracy of the output of the target virtual application object. .
  • an embodiment of the present application provides a storage medium in which multiple instructions are stored, and the instructions can be loaded by a processor to execute the steps in any virtual application object output method provided in the embodiments of the present application.
  • the instruction can perform the following steps:
  • the current state information is used to indicate that the virtual application object is in a known state or an unknown state.
  • a virtual application object state plane is constructed, Among them, the virtual application object state plane includes the area corresponding to each virtual application object, and the area includes the current state information of its corresponding virtual application object.
  • the corresponding output of the multiple virtual application objects to be output is determined. Probability, according to the respective output probabilities of the virtual application objects to be output, determine the target virtual application object from the multiple virtual application objects to be output for output.
  • the storage medium may include: read only memory (ROM, Read Only Memory), random access memory (RAM, Random Access Memory), magnetic disk or optical disk, etc.
  • any virtual application object output method provided in the embodiments of the present application can be implemented.
  • the beneficial effects that can be achieved are detailed in the previous embodiments, and will not be repeated here.

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Human Computer Interaction (AREA)
  • User Interface Of Digital Computer (AREA)
  • Processing Or Creating Images (AREA)

Abstract

一种虚拟应用对象输出方法、装置以及计算机存储介质，所述方法包括：获取虚拟应用中多个虚拟应用对象的当前状态信息（S201），当前状态信息用于指示虚拟应用对象处于已知状态或者处于未知状态；根据多个虚拟应用对象的当前状态信息，构建虚拟应用对象状态平面（S202），其中，虚拟应用对象状态平面包括每个虚拟应用对象对应的区域，区域中包括虚拟应用对象的当前状态信息；根据虚拟应用对象状态平面，确定多个待输出虚拟应用对象各自对应的输出概率（S203）；根据多个待输出虚拟应用对象各自对应的输出概率，从所述多个待输出虚拟应用对象中确定目标虚拟应用对象进行输出（S204）。上述方法可以提高目标虚拟应用对象输出的准确性。

Description

一种虚拟应用对象输出方法、装置以及计算机存储介质
本申请要求于2019年09月26日提交中国专利局、申请号为201910914842.0、申请名称为“一种虚拟应用对象输出方法、装置以及计算机存储介质”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
本申请涉及人工智能(Artificial Intelligence，AI)领域，具体涉及虚拟应用对象处理技术。
背景技术
网络游戏通常以互联网为传输媒介，以游戏运营商服务器和用户计算机为处理终端，以游戏客户端软件为信息交互窗口，其包括旨在实现娱乐、休闲、交流、以及取得虚拟成就的具有可持续性的个体性多人在线游戏。
随着网络游戏的发展，用户对于网络游戏的要求也在不断提高，对于包括虚拟应用对象输出步骤的网络游戏，如能够输出推荐操作的棋牌类游戏等，用户希望能够输出准确的虚拟应用对象，然而，对于一些网络游戏，由于其中虚拟应用对象的种类繁多，因此虚拟应用对象输出的准确性往往不够高。
发明内容
本申请实施例提供一种虚拟应用对象输出方法、装置以及计算机存储介质，可以提高目标虚拟应用对象输出的准确性。
本申请实施例提供一种虚拟应用对象输出方法，由网络设备执行，包括：
获取虚拟应用中多个虚拟应用对象的当前状态信息，所述当前状态信息用于指示虚拟应用对象处于已知状态或者处于未知状态；
根据所述多个虚拟应用对象的当前状态信息，构建虚拟应用对象状态平面，其中，所述虚拟应用对象状态平面包括每个虚拟应用对象对应的区域，所述区域中包括其对应的所述虚拟应用对象的当前状态信息；
根据所述虚拟应用对象状态平面，确定多个待输出虚拟应用对象各自对应的输出概率；
根据所述多个待输出虚拟应用对象各自对应的输出概率，从多个待输出虚拟应用对象中确定目标虚拟应用对象进行输出。
相应的，本申请实施例还提供一种虚拟应用对象输出装置，包括：
获取模块，用于获取虚拟应用中多个虚拟应用对象的当前状态信息，所述当前状态信息用于指示虚拟应用对象处于已知状态或者处于未知状态；
构建模块，用于根据所述多个虚拟应用对象的当前状态信息，构建虚拟应用对象状态平面，其中，所述虚拟应用对象状态平面包括每个虚拟应用对象对应的区域，所述区域按照预设排列规则进行排列，所述区域中包括其对 应的所述虚拟应用对象的当前状态信息；
概率确定模块，用于根据所述虚拟应用对象状态平面，确定多个待输出虚拟应用对象各自对应的输出概率；
输出模块，用于根据所述多个待输出虚拟应用对象各自对应的输出概率，从多个待输出虚拟应用对象中确定目标虚拟应用对象进行输出。
此外，本申请实施例还提供一种存储介质，所述存储介质存储有多条指令，所述指令适于处理器进行加载，以执行本申请实施例提供的任一种虚拟应用对象输出方法中的步骤。
此外，本申请实施例还提供一种计算机程序产品，包括指令，当其在计算机上运行时，使得计算机执行实施例提供的任一种虚拟应用对象输出方法中的步骤。
本申请实施例可以获取虚拟应用中多个虚拟应用对象的当前状态信息，当前状态信息用于指示虚拟应用对象处于已知状态或者处于未知状态，根据多个虚拟应用对象的当前状态信息，构建虚拟应用对象状态平面，其中，虚拟应用对象状态平面包括每个虚拟应用对象对应的区域，区域中包括其对应的虚拟应用对象的当前状态信息，根据虚拟应用对象状态平面，确定多个待输出虚拟应用对象各自对应的输出概率，根据多个待输出虚拟应用对象各自对应的输出概率，从多个待输出虚拟应用对象中确定目标虚拟应用对象进行输出。该方案通过将虚拟应用中的虚拟应用对象的当前状态信息用状态平面的方式进行表示，使得虚拟应用对象的当前状态信息能够简明准确的表示在状态平面中，从而方便准确地确定待输出虚拟应用对象对应的输出概率，有助于提高目标虚拟应用对象输出的准确性。
附图说明
图1是本申请实施例提供的虚拟应用对象输出系统的场景示意图；
图2是本申请实施例提供的虚拟应用对象输出方法的第一流程图；
图3是本申请实施例提供的虚拟应用对象输出方法的第二流程图；
图4是本申请实施例提供的虚拟应用对象输出方法的第三流程图；
图5是本申请实施例提供的虚拟应用对象输出方法的第四流程图；
图6是本申请实施例提供的虚拟应用对象输出方法的第五流程图；
图7是本申请实施例提供的网络麻将游戏应用第一界面的示意图；
图8是本申请实施例提供的网络麻将游戏应用第二界面的示意图；
图9是本申请实施例提供的概率获取网络的训练流程示意图；
图10是本申请实施例提供的应用概率获取网络输出目标虚拟应用对象的流程示意图；
图11是本申请实施例提供的第一种虚拟应用对象状态平面的第一示意图；
图12是本申请实施例提供的第一种虚拟应用对象状态平面的第二示意 图；
图13是本申请实施例提供的第一种虚拟应用对象状态平面的第三示意图；
图14是本申请实施例提供的第一种虚拟应用对象状态平面的第四示意图；
图15是本申请实施例提供的第二种虚拟应用对象状态平面的第一示意图；
图16是本申请实施例提供的第二种虚拟应用对象状态平面的第二示意图；
图17是本申请实施例提供的第二种虚拟应用对象状态平面的第三示意图；
图18是本申请实施例提供的第三种虚拟应用对象状态平面的第一示意图；
图19是本申请实施例提供的第三种虚拟应用对象状态平面的第二示意图；
图20是本申请实施例提供的第三种虚拟应用对象状态平面的第三示意图；
图21a是本申请实施例提供的第四种虚拟应用对象状态平面的示意图；
图21b是本申请实施例提供的第五种虚拟应用对象状态平面的示意图；
图22是本申请实施例提供的虚拟应用对象输出装置的结构示意图；
图23是本申请实施例提供的网络设备的结构示意图。
具体实施方式
本申请实施例提供一种虚拟应用对象输出方法、装置、以及计算机存储介质。其中，该虚拟应用对象输出装置可以集成在网络设备中，该网络设备可以为终端、服务器等设备；其中，终端可以为手机、平板电脑、笔记本电脑、个人计算机(PC，Personal Computer)、微型处理盒子等设备；服务器可以为应用服务器或web服务器，具体部署时，可以为独立的服务器，也可以为集群服务器，还可以为云服务器。
请参阅图1，图1为本申请实施例提供的虚拟应用对象输出方法的应用场景示意图，以虚拟应用对象输出装置集成在网络设备中为例，网络设备可以获取虚拟应用中多个虚拟应用对象的当前状态信息，当前状态信息用于指示虚拟应用对象处于已知状态或者处于未知状态，根据多个虚拟应用对象的当前状态信息，构建虚拟应用对象状态平面，其中，虚拟应用对象状态平面包括每个虚拟应用对象对应的区域，区域中包括与其对应的虚拟应用对象的当前状态信息，根据虚拟应用对象状态平面，确定多个待输出虚拟应用对象各自对应的输出概率，根据多个待输出虚拟应用对象各自对应的输出概率，从多个待输出虚拟应用对象中确定目标虚拟应用对象进行输出。
以下分别进行详细说明。需说明的是，以下实施例的描述顺序不作为对实施例优选顺序的限定。
本发明实施例提供的一种虚拟应用对象输出方法，下面以该方法由服务器执行为例进行介绍，如图2所示，该虚拟应用对象输出方法的具体流程可以如下：
201、获取虚拟应用中多个虚拟应用对象的当前状态信息。
其中，虚拟应用可以为安装在终端上的应用软件，能够满足用户对于不同领域、不同问题的应用需求，并且为用户提供丰富的使用体验。比如，本申请实施例中的虚拟应用可以为游戏应用，该游戏应用可以为将各种程序和动画效果结合得到的软件产品，通过游戏应用，可以对用户的脑、眼、手等器官进行锻炼，提升用户的逻辑力、敏捷力等等。又比如，本申请实施例中的虚拟应用还可以为需要向用户提供推荐操作的牌类游戏应用，如网络麻将游戏应用、网络扑克游戏应用等等。
其中，虚拟应用对象可以为虚拟应用中与应用相关的虚拟对象，比如，当虚拟应用为游戏应用时，该虚拟应用对象可以为游戏对象，如在网络麻将游戏应用中，一个虚拟应用对象可以为一张虚拟麻将牌；在网络牌类游戏应用中，一个虚拟应用对象可以为一张虚拟牌，等等。
其中，当前状态信息可以为指示虚拟应用中虚拟应用对象当前状态的状态信息，比如，当虚拟应用为网络麻将游戏应用时，当前状态信息可以指示虚拟麻将牌当前的状态信息，如当前状态信息可以用于指示虚拟麻将牌当前处于已知状态还是未知状态。
比如，当虚拟应用为网络麻将游戏应用时，在当前玩家的视角，对于当前玩家自己已知的虚拟麻将牌、各个玩家已经输出的虚拟麻将牌、以及其他玩家展示出来的虚拟麻将牌来说，其对应的当前状态信息应指示虚拟麻将牌处于已知状态；而对于其余无法知晓具体状态的虚拟麻将牌来说，其对应的当前状态信息应指示虚拟麻将牌处于未知状态。
其中，虚拟应用对象的当前状态信息的定义方法，还可以根据实际情况进行调整，如在当前玩家的视角，还可以将当前玩家自己已知的虚拟麻将牌、各个玩家已经输出的虚拟麻将牌、其他玩家展示出来的虚拟麻将牌、以及无法得知具体状态的虚拟麻将牌，分别识别为不同种类的当前状态信息，等等。
其中，本申请实施例不对当前状态信息的定义方式进行限定，当前状态信息的定义方式可以根据虚拟应用的不同、虚拟应用对象的不同、以及虚拟应用中游戏规则的不同等进行适当调整。通过灵活地对虚拟应用对象的当前状态信息进行定义，可以提升该虚拟应用对象输出方法的灵活性，使其能够适应更多种类的虚拟应用。
其中，当前状态信息还可以通过标识的形式进行表示，当前状态信息的表示方法可以有多种，比如，可以通过不同的颜色指示不同种类的当前状态 信息；或者可以通过不同的图案、字符等指示不同种类的当前状态信息，等等。比如，可以通过二值化标识“0”或“1”表示当前状态信息，在当前状态信息为“0”时，可以指示对应的虚拟麻将牌处于未知状态；在当前状态信息为“1”时，可以指示对应的虚拟麻将牌处于已知状态，等等。
在实际应用中，比如，当虚拟应用为网络麻将游戏应用，虚拟应用对象为虚拟麻将牌时，可以确定每张虚拟麻将牌对应的当前状态信息。其中，在当前玩家的视角，可以将当前玩家自己已知的虚拟麻将牌、各个玩家已经输出的虚拟麻将牌、以及其他玩家展示出来的虚拟麻将牌等已知状态的虚拟麻将牌识别为已知状态，并将已知状态虚拟麻将牌的当前状态信息记录为“1”；而其余无法知晓具体状态的虚拟麻将牌可以识别为未知状态，并将未知状态虚拟麻将牌的当前状态信息记录为“0”。通过将虚拟麻将牌的状态区分为已知与未知，不仅简化了虚拟麻将牌的状态识别算法，而且将状态区分为两种，可以方便后续针对输出虚拟应用对象的工作。
在一实施例中，又比如，还可以将当前玩家自己已知的虚拟麻将牌、各个玩家已经输出的虚拟麻将牌、其他玩家展示出来的虚拟麻将牌、以及其余无法知晓具体状态的虚拟麻将牌分别通过不同的颜色或者不同的字符进行标记，通过不同的颜色或者不同的字符，表示不同种类的当前状态信息，等等。
通过多种方法进行当前状态信息的定义，可以提升本申请实施例中虚拟应用对象输出方法的灵活性，使其能够适应更多种类的虚拟应用，进而提升虚拟应用对象输出的准确性。
202、根据多个虚拟应用对象的当前状态信息，构建虚拟应用对象状态平面。
其中，虚拟应用对象状态平面可以为表征虚拟应用中所有虚拟应用对象的当前状态的平面，通过该平面可以得知所有虚拟应用对象的当前状态，并且可以基于该平面进行计算，从而确定输出的目标虚拟应用对象。比如，如图11所示，该虚拟应用对象状态平面中可以包括每个虚拟应用对象对应的区域，这些区域按照预设排列规则进行排列，每个区域用于记录该区域对应的虚拟应用对象的当前状态信息。
在实际应用中，由于某些牌类游戏中的牌会包括不同的花色、不同的牌型、以及花色与牌型的组合，如麻将中的麻将牌包括万、条、筒、风、箭等不同的花色，还包括一、二、三等不同的牌面值，花色与牌面值进行组合后会组合成34种不同的麻将牌。因此，可以通过将麻将牌的当前状态信息以二值化的形式在平面上进行表示，以便于后续网络模型的学习。
比如，当虚拟应用为网络麻将游戏应用，虚拟应用对象为虚拟麻将牌时，获取到每张虚拟麻将牌的当前状态信息后，可以根据每张虚拟麻将牌的当前状态信息，构建出虚拟应用对象状态平面。如图11所示，虚拟应用对象状 态平面中包括多个矩形区域，每个矩形区域都代表一张虚拟麻将牌。多个矩形区域根据每个区域对应的虚拟麻将牌的麻将牌名称，按照“一万、二万、三万、四万、五万、六万、七万、八万、九万、一条、二条、三条、四条、五条、六条、七条、八条、九条、一筒、二筒、三筒、四筒、五筒、六筒、七筒、八筒、九筒、东、南、西、北、中、发、白”的顺序进行排列，整体排列为一个4*34的矩形阵列，其中矩形阵列中每行都包括34个名称不同的虚拟麻将牌，矩形阵列中每列都包括4个名称相同的虚拟麻将牌。每个矩形区域中都包括其对应的虚拟麻将牌的当前状态信息“0”或“1”，其中，包括“0”的矩形区域可以指示该矩形区域对应的虚拟麻将牌的状态未知，包括“1”的矩形区域可以指示该矩形区域对应的虚拟麻将牌的状态已知。其中，通过这种方法构建出的虚拟应用对象状态平面为一种二值平面。
在一实施例中，比如，为了更加明显地在虚拟应用对象状态平面中，对未知状态的虚拟麻将牌和已知状态的虚拟麻将牌进行区分，如图12所示，还可以通过不同的颜色表示不同的当前状态信息。
在一实施例中，比如，由于麻将游戏中，和牌方式有多种，因此为了提升该虚拟应用对象输出方法的灵活性，还可以对矩形区域的预设排列规则进行调整，比如，还可以将花色为“条”或者“筒”的虚拟麻将牌对应的区域设置于平面最左端；或者在平面里同花色对应的区域中，按照虚拟麻将牌的牌面值从大到小的方式进行从左到右的排列，等等。
其中，番种可以为麻将中对具有一定分值的各种牌张组合形式或者和牌方式的名称。当牌型条件符合规定，并且达到或者超过和分标准，此时，可以认为玩家和牌。
在一实施例中，比如，还可以从该虚拟应用对象状态平面中找到多种可能的番种子模式，如图13所示，图中粗线包络区域中的虚拟麻将牌分别可以构成“一色三步高”、“三色三同刻”、以及“大四喜”的番种，通过这种方式进行虚拟麻将牌状态平面的表示，使得计算机能够根据该状态平面识别出多种番种，进而进行虚拟应用对象输出的预测。
在一实施例中，按照上述方法构建的虚拟应用对象状态平面可能会存在番种误判的情况，如图14所示，图中的区域一可以表示番种“一色三步高”，区域二虽然区域形状与区域一相同，但是并非番种“一色三步高”的表示形式，这样会对后续的模型学习造成一定的困难。
因此，可以通过在虚拟应用对象状态平面中添加隔离区域的方式，将不同类型的虚拟应用对象在状态平面中进行严格区分，以解决番种误判的问题。具体地，步骤“根据所述多个虚拟应用对象的当前状态信息，构建虚拟应用对象状态平面”，可以包括：
获取多个虚拟应用对象的对象类型；
根据所述当前状态信息、以及所述对象类型，构建虚拟应用对象状态平 面，其中，所述虚拟应用对象状态平面包括若干状态子平面、以及若干隔离区域，所述状态子平面与所述对象类型相对应，所述隔离区域位于相邻的两个状态子平面之间。
在实际应用中，比如，当虚拟应用为网络麻将游戏应用，虚拟应用对象为虚拟麻将牌时，可以将虚拟麻将牌划分为多个对象类型，如可以将虚拟麻将牌划分为“万”、“条”、“筒”、“风”、以及“箭”五种对象类型，其中，对象类型“万”可以包括“一万”、“二万”、“三万”、“四万”、“五万”、“六万”、“七万”、“八万”、以及“九万”九种名称的虚拟麻将牌；对象类型“条”可以包括“一条”、“二条”、“三条”、“四条”、“五条”、“六条”、“七条”、“八条”、以及“九条”九种名称的虚拟麻将牌；对象类型“筒”可以包括“一筒”、“二筒”、“三筒”、“四筒”、“五筒”、“六筒”、“七筒”、“八筒”、以及“九筒”九种名称的虚拟麻将牌；对象类型“风”可以包括“东”、“南”、“西”、以及“北”四种名称的虚拟麻将牌；对象类型“箭”可以包括“中”、“发”、以及“白”三种名称的虚拟麻将牌。
获取到每张虚拟麻将牌对应的对象类型后，可以根据每张虚拟麻将牌的当前状态信息、以及每张虚拟麻将牌对应的对象类型，构建虚拟应用对象状态平面，如图15所示，该虚拟应用对象状态平面包括5个状态子平面、以及4个隔离区域，其中，5个状态子平面分别对应“万”、“条”、“筒”、“风”、以及“箭”五种对象类型。隔离区域为0值列，由4个矩形区域构成，每个矩形区域中表示的当前状态信息都为0，并且隔离区域位于相邻的两个状态子平面之间。
其中，该虚拟应用对象状态平面为4*38的矩形阵列，从左向右数第10列、第20列、第30列、以及第35列的区域都为隔离区域。从左向右数第1列至第9列的状态子平面对应对象类型“万”，第11列至第19列的状态子平面对应对象类型“条”，第21列至第29列的状态子平面对应对象类型“筒”，第31列至第34列的状态子平面对应对象类型“风”，第36列至第38列的状态子平面对应对象类型“箭”。
此时，如图16所示，虚拟应用对象状态平面中区域一可以表示番种“一色三步高”，而区域二由于隔离区域的隔离作用，与区域一的形状不相同，从而减少番种误判的可能性。
在一实施例中，如图15所示的包括隔离区域的虚拟应用对象状态平面，还可以通过对如图12所示的不包括隔离区域的状态平面进行变换得到。具体地，步骤“根据所述当前状态信息、以及所述对象类型，构建虚拟应用对象状态平面”，可以包括：
根据所述当前状态信息，构建第一初始状态平面；
根据所述对象类型，将所述第一初始状态平面分割为若干状态子平面；
在两个相邻的状态子平面之间插入隔离区域，得到虚拟应用对象状态平面。
在实际应用中，比如，可以首先根据虚拟麻将牌的当前状态信息，构建出如图12所示的第一初始状态平面，然后，将第一初始状态平面按照“万”、“条”、“筒”、“风”、以及“箭”五种对象类型，分割为五个状态子平面，每个状态子平面都对应一种对象类型，然后在相邻的两个状态子平面中插入一列隔离区域，得到如图15所示的虚拟应用对象状态平面。
在一实施例中，构建如图15所示的虚拟应用对象状态平面的方法有多种，不限于上述给出的方法。
在一实施例中，按照上述方法构建的虚拟应用对象状态平面依然会存在番种误判的情况，如图17所示，图中的区域三可以表示番种“一色三同顺”，其中，区域四虽然区域形状与区域三相同，但并非番种“一色三同顺”的表示形式，而是番种“三风刻”的表示形式，区域五虽然区域形状与区域三相同，但并非番种“一色三同顺”的表示形式，而是番种“大三元”的表示形式，这样会对后续的模型学习造成一定的困难。
因此，可以通过在虚拟应用对象状态平面中，对需要进行隔离的状态子平面之间添加隔离区域的方式，将需要进行隔离的虚拟应用对象在状态平面中进行更加严格的区分，以解决番种误判的问题。具体地，步骤“根据所述当前状态信息、以及所述对象类型，构建虚拟应用对象状态平面”，可以包括：
从若干对象类型中，确定需要进行隔离的目标对象类型，所述目标对象类型包括多个目标对象子类型；
根据所述当前状态信息、所述对象类型、以及所述目标对象子类型，构建虚拟应用对象状态平面，其中，所述虚拟应用对象状态平面中的状态子平面包括若干隔离后状态区域，所述隔离后状态区域与所述目标对象子类型相对应，所述虚拟应用对象状态平面中的隔离区域位于相邻的两个隔离后状态区域之间。
在实际应用中，比如，当虚拟应用为网络麻将游戏应用，虚拟应用对象为虚拟麻将牌时，可以从“万”、“条”、“筒”、“风”、以及“箭”五种对象类型中，确定需要进行隔离的“风”和“箭”两种目标对象类型，并将目标对象类型“风”划分为“东”、“南”、“西”、以及“北”四种目标对象子类型；将目标对象类型“箭”划分为“中”、“发”、以及“白”三种目标对象子类型。
然后根据当前状态信息、对象类型、以及目标对象子类型，构建虚拟应用对象状态平面。如图18所示，该虚拟应用对象状态平面包括3个状态子平面、7个隔离后状态区域、以及9个隔离区域，其中，3个状态子平面分别对应“万”、“条”、以及“筒”三种对象类型，7个隔离后状态区域分 别对应“东”、“南”、“西”、“北”、“中”、“发”、“白”七种目标对象子类型。
其中，隔离区域为0值列，分别位于相邻的对象类型对应的状态子平面之间、相邻的目标对象子类型对应的隔离后状态区域之间、以及相邻的状态子平面和隔离后状态区域之间。
其中，该虚拟应用对象状态平面为4*46的矩形阵列，从左向右数第10列、第20列、第30列、第32列、第34列、第36列、第38列、第39列、第41列、第42列、第44列、第45列的区域都为隔离区域。从左向右数第1列至第9列的状态子平面对应对象类型“万”，第11列至第19列的状态子平面对应对象类型“条”，第21列至第29列的状态子平面对应对象类型“筒”，第31列的隔离后状态区域对应目标对象子类型“东”，第33列的隔离后状态区域对应目标对象子类型“南”，第35列的隔离后状态区域对应目标对象子类型“西”，第37列的隔离后状态区域对应目标对象子类型“北”，第40列的隔离后状态区域对应目标对象子类型“中”，第43列的隔离后状态区域对应目标对象子类型“发”，第46列的隔离后状态区域对应目标对象子类型“白”。
在一实施例中，由于对象类型“风”、以及对象类型“箭”需要在平面中进行严格的区分，因此，可以通过不同区域尺寸的隔离区域进行状态子平面之间的隔离，达到严格区分类型的目的。具体地，步骤“根据所述当前状态信息、所述对象类型、以及所述目标对象子类型，构建虚拟应用对象状态平面”，可以包括：
确定需要进行隔离的两个目标对象子类型之间的隔离区域的区域尺寸；
根据所述当前状态信息、所述对象类型、所述目标对象子类型、以及所述区域尺寸，构建虚拟应用对象状态平面，其中，所述虚拟应用对象状态平面中的两个隔离后状态区域之间包括区域尺寸的隔离区域。
在实际应用中，比如，可以确定目标对象子类型“东”与“南”之间的隔离区域的区域尺寸为1，目标对象子类型“北”与“中”之间的隔离区域的区域尺寸为2，目标对象子类型“中”与“发”之间的隔离区域的区域尺寸为2，等等。区域尺寸都确定完毕后，可以根据当前状态信息、对象类型、目标对象子类型、以及区域尺寸，构建虚拟应用对象状态平面。如图18所示，“东”对应的隔离后状态区域与“南”对应的隔离后状态区域之间的隔离区域的列数为1，而“中”对应的隔离后状态区域与“发”对应的隔离后状态区域之间的隔离区域的列数为2。此时，对对象类型“风”、以及对象类型“箭”进行了严格的区分。
其中，在此实施例中不对隔离区域的列数作过多的限定，只要保证在需要进行严格区分的目标对象子类型对应的隔离后状态区域之间，隔离区域的列数不同即可。
此时，如图19所示，虚拟应用对象状态平面中区域三可以表示番种“一色三同顺”，区域四由于不同列数的隔离区域的隔离作用，与区域三的形状不相同，可以表示番种“三风刻”，区域五由于不同列数的隔离区域的隔离作用，与区域三和区域四的形状都不相同，可以表示番种“大三元”，从而降低了番种误判的可能性。
在一实施例中，如图18所示的包括列数不完全相同的隔离区域的虚拟应用对象状态平面，还可以通过对如图15所示的包括列数相同的隔离区域的状态平面进行变换得到。具体地，步骤“根据所述当前状态信息、所述对象类型、所述目标对象子类型、以及所述区域尺寸，构建虚拟应用对象状态平面”，可以包括：
根据所述当前状态信息、以及所述对象类型，构建第二初始状态平面；
确定所述第二初始状态平面中所述目标对象类型对应的目标状态子平面；
根据所述目标对象子类型，将所述目标状态子平面分割为若干隔离后状态区域；
在两个相邻的隔离后状态区域之间插入所述区域尺寸的隔离区域，得到隔离后初始状态平面；
根据所述隔离后初始状态平面、以及所述第二初始状态平面，构建虚拟应用对象状态平面。
在实际应用中，比如，可以根据当前状态信息、以及对象类型，构建如图15所示的第二初始状态平面，然后确定目标对象类型“风”和目标对象类型“箭”对应的目标状态子平面，然后将目标对象类型“风”对应的目标状态子平面分割为四个隔离后状态区域，分别对应“东”、“南”、“西”、以及“北”四种目标对象子类型；将目标对象类型“箭”对应的目标状态子平面分割为三个隔离后状态区域，分别对应“中”、“发”、以及“白”三种目标对象子类型。然后在“东”、“南”、“西”、以及“北”四种目标对象子类型对应的隔离后状态区域之间插入一列隔离区域，在“中”、“发”、以及“白”三种目标对象子类型对应的隔离后状态区域之间插入两列隔离区域，在“北”、以及“中”对应的隔离后状态区域之间插入两列隔离区域，得到隔离后初始状态平面。最后将第二初始状态平面中“风”和“箭”对应的目标状态子平面，替换为隔离后初始状态平面，得到虚拟应用对象状态平面。
在一实施例中，构建如图18所示的虚拟应用对象状态平面的方法有多种，不限于上述给出的方法。
在一实施例中，由于矩形平面的长宽差距越小，越适合神经网络进行学习，因此，还可以通过对平面进行旋转的方式，使其长宽差距变小。比如，如图21a所示，图中虚拟应用对象状态平面中包括5个旋转后状态子平面、 以及多个隔离区域，每个旋转后状态子平面对应一种对象类型，隔离区域将不同的对象类型、以及不同的目标对象子类型隔离开。图中虚拟应用对象状态平面为9*24的矩形阵列，纵向方向表示虚拟麻将牌的牌面值1至9，横向方向分别为对应“万”、“条”、“筒”、“风”、以及“箭”的5个旋转后状态子平面。应理解，对平面进行旋转处理后，也可以得到如图21b所示的状态平面，即得到24*9的矩形阵列。
在一实施例中，如图21a和图21b所示的虚拟应用对象状态平面还可以通过图18所示的状态平面变换得到。具体地，步骤“根据所述当前状态信息、所述对象类型、所述目标对象子类型、以及所述区域尺寸，构建虚拟应用对象状态平面”，可以包括：
根据所述当前状态信息、所述对象类型、所述目标对象子类型、以及所述区域尺寸，构建第三初始状态平面；
根据所述对象类型，将所述第三初始状态平面分割为若干待旋转状态子平面；
将每个待旋转状态子平面进行旋转，得到旋转后状态子平面；
在相邻两个旋转后状态子平面之间插入隔离区域，得到虚拟应用对象状态平面。
在实际应用中，比如，可以根据当前状态信息、对象类型、目标对象子类型、以及区域尺寸，构建如图20所示的第三初始状态平面，然后根据“万”、“条”、“筒”、“风”、以及“箭”五种对象类型，将第三初始状态平面分割为5个待旋转状态子平面、以及3个隔离区域。然后将每个待旋转状态子平面进行旋转，得到5个旋转后状态子平面，并将“箭”对应的旋转后子平面上添加2*4的隔离区域，使得5个旋转后状态子平面可以拼接为矩形，然后在相邻两个旋转后状态子平面中插入1列隔离区域，得到虚拟应用对象状态平面。
在一实施例中，构建如图21a和21b所示的虚拟应用对象状态平面的方法有多种，不限于上述给出的方法。
如图21a和21b所示的虚拟应用对象状态平面中，隔离区域分别从宽度方向和高度方向进行插入，减少状态平面中多种番种子模式发生混淆的可能性，从而降低番种误判的可能性。并且整个状态平面更接近于正方形，有利于神经网络进行学习。
203、根据虚拟应用对象状态平面，确定多个待输出虚拟应用对象各自对应的输出概率。
其中，待输出虚拟应用对象可以为根据游戏规则设置，能够进行输出的虚拟应用对象。比如，当虚拟应用为网络麻将游戏应用，虚拟应用对象为虚拟麻将牌时，待输出虚拟应用对象可以为当前玩家能够打出的虚拟麻将牌。
在实际应用中，比如，构建虚拟应用对象状态平面完毕后，可以根据该 虚拟应用对象状态平面，进行每个待输出虚拟应用对象对应的输出概率的确定，如图7所示，确定出每张待输出虚拟应用对象对应的输出概率后，还可以在游戏界面上以表格的形式进行展示，如玩家北风输出待输出虚拟应用对象“南”的概率为62.54％，玩家北风输出待输出虚拟应用对象“中”的概率为25％，等等。
在一实施例中，如图7所示，可以通过列表的形式在游戏界面上显示每一张待输出虚拟应用对象对应的输出概率，还可以将概率过低的待输出虚拟应用对象进行省略，等等，以提高游戏界面显示的灵活性。
在一实施例中，为了提升计算效率以及准确性，还可以通过神经网络对虚拟应用对象状态平面进行计算。具体地，步骤“根据所述虚拟应用对象状态平面，确定多个待输出虚拟应用对象各自对应的输出概率”，可以包括：
将所述虚拟应用对象状态平面输入至概率获取网络；
基于所述概率获取网络，获取待输出虚拟应用对象对应的输出概率。
其中，概率获取网络可以为神经网络，神经网络可以为通过模仿动物神经网络行为特征，进行分布式并行信息处理的算法数学模型。神经网络依靠系统的复杂程度，通过调整内部大量节点之间相互连接的关系，达到处理信息的目的。
在一实施例中，将以概率获取网络为CNN(Convolutional Neural Networks，卷积神经网络)模型为例进行说明。卷积神经网络模型的基本组件包括卷积层、池化层、全连接层等等，其中，卷积层、池化层等可组成一个卷积块，而多个卷积块连接多个全连接层形成了卷积神经网络结构。但在本申请的其他实施例中，该概率获取网络也可以为如RNN(Recurrent Neural Network，循环神经网络)、DNN(Deep Neural Network，深度神经网络)、随机森林模型、SVM(Support Vector Machine，支持向量机)等其他多分类模型框架，且本实施例中并不以此为限。
在实际应用中，比如，可以将虚拟应用对象状态平面输入至CNN(卷积神经网络，Convolutional Neural Networks)中，该卷积神经网络可以称为概率获取网络，经过概率获取网络的计算，可以计算得到当前玩家能够支配输出的待输出虚拟应用对象的输出概率。
在一实施例中，可以将虚拟应用对象状态平面视为像素值的矩阵，由于虚拟应用对象状态平面中仅包括0和1两种当前状态信息，因此，该虚拟应用对象状态平面为一种二值图像。
其中，概率获取网络中可以包括若干卷积层，通过卷积层可以对输入的图像进行卷积操作。卷积操作可以从输入的图像数据中学习到图像的特征，并且保留像素间的空间关系。在卷积操作中，卷积核以一定的步长在输入的图像中进行移动，并进行卷积运算，然后可以输出图像对应的特征。因此，将虚拟应用对象状态平面输入至概率获取网络中后，可以通过概率获取网络 中的卷积层，对虚拟应用对象状态平面进行特征提取。
提取出虚拟应用对象状态平面对应的特征后，可以通过全连接层确定待输出虚拟应用对象对应的输出概率，全连接层可以将学到的特征映射到样本标记空间，在网络模型中起到“分类器”的作用。全连接层的每一个结点都与上一层输出的结点相连，其中，全连接层的一个结点即称为全连接层中的一个神经元，全连接层中神经元的数量可以根据实际应用的需求而定。全连接的核心操作为矩阵向量乘积，本质上是由一个特征空间线性变换到另一个特征空间。在概率获取网络中，全连接层可以位于最后几层，用于对前面获取的特征作加权和。
在一实施例中，为了提升概率获取网络的网络性能，全连接层每个神经元还可以采用ReLU函数。最后一层全连接层的输出值被传递给一个输出，可以采用softmax层进行分类，经过分类后，可以计算得到当前玩家能够支配输出的各个待输出虚拟应用对象各自对应的输出概率。
在一实施例中，为了提升概率获取网络确定输出概率的准确性，还可以对概率获取网络进行训练。具体地，该虚拟应用对象输出方法，还可以包括：
获取若干样本状态平面、以及每个样本状态平面对应的真实输出对象；
根据所述样本状态平面对预设概率获取网络进行训练，得到所述样本状态平面对应的预测输出对象；
根据所述样本状态平面对应的真实输出对象和所述预测输出对象，对预设概率获取网络进行收敛，得到概率获取网络。
在实际应用中，比如，如图9所示，可以获取之前存储的游戏日志，并对获取到的游戏日志进行整理，其中，游戏日志中可以记录游戏从开局到结束的过程中，全部玩家进行的全部游戏动作，可以将游戏日志中的信息整理为多张样本状态图像。并且从多个游戏动作中确定出若干重要的目标游戏动作，该目标游戏动作可以作为真实输出对象，如当前玩家输出“中”的游戏动作可以识别为目标游戏动作。然后获取该目标游戏动作之前的多张样本状态图像，如可以获取目标游戏动作之前的100张样本状态图像，此时，这100张样本状态图像都与该目标游戏动作相对应，也即一个真实输出对象可以对应多张样本状态平面。其中，该样本状态平面的表示形式可以与虚拟应用对象状态平面的表示形式相同。
然后，可以根据获取到的多张样本状态平面对预设概率获取网络进行训练，得到样本状态平面对应的预测输出对象，并根据该样本状态平面对应的真实输出对象和预测输出对象，对预设概率获取网络的模型参数进行收敛，当计算得出的预测输出对象与真实输出对象之间的匹配度满足预设阈值时，说明此时的概率获取网络已经训练完毕，可以得到训练好的概率获取网络。
204、根据多个待输出虚拟应用对象各自对应的输出概率，从所述多个待输出虚拟应用对象中确定目标虚拟应用对象进行输出。
在实际应用中，比如，概率获取网络可以输出每个待输出虚拟应用对象对应的输出概率，然后，可以根据获取到的概率分布进行采样，从多个待输出虚拟应用对象中确定目标虚拟应用对象，并将该目标虚拟应用对象进行输出。
在一实施例中，网络设备对目标虚拟应用对象进行输出的方式有多种，比如，当该虚拟应用对象输出方法应用于计算机侧的虚拟玩家时，可以直接帮助计算机侧的虚拟玩家输出该目标虚拟应用对象，从而推进游戏进程。
又比如，当该虚拟应用对象输出方法应用于用户终端的真实玩家时，确定目标虚拟应用对象后，可以将游戏界面中该目标虚拟应用对象对应的虚拟麻将牌进行突出显示，如可以使该虚拟麻将牌进行闪烁或者用不同的颜色对该虚拟麻将牌进行标记，此时，用户可以得知输出这张虚拟麻将牌能够推进游戏进程，用户可以选择输出进行特殊标记的虚拟麻将牌，当然，用户也可以出于其他的考虑选择其他的虚拟麻将牌进行输出，从而提升了游戏的灵活性。
在一实施例中，输出了目标虚拟应用对象后，游戏可能还未结束，此时，可以对当前状态信息进行更新，并重新进行目标虚拟应用对象的确定。具体地，在步骤“根据所述多个待输出虚拟应用对象各自对应的输出概率，从所述多个待输出虚拟应用对象中确定目标虚拟应用对象进行输出”之后，还可以包括：
更新待输出虚拟应用对象；
当所述待输出虚拟应用对象不满足终止条件时，更新虚拟应用对象的当前状态信息；
返回执行根据所述多个虚拟应用对象的当前状态信息，构建虚拟应用对象状态平面的步骤；
当所述待输出虚拟应用对象满足终止条件时，终止输出。
在实际应用中，比如，当虚拟应用为网络麻将游戏应用，虚拟应用对象为虚拟麻将牌时，输出了目标虚拟应用对象后，虚拟应用中的多个待输出虚拟应用对象发生了变化，也即，已经打出的牌不能够进行再次输出，此时，可以对待输出虚拟应用对象进行更新。然后对更新后的待输出虚拟应用对象进行分析，如可以对此时当前玩家能够输出的多张待输出虚拟应用对象进行分析，当这些待输出虚拟应用对象不能够组成和牌的牌型时，可以认为待输出虚拟应用对象不满足终止条件，此时可以返回执行根据多个虚拟应用对象的当前状态信息，构建虚拟应用对象状态平面的步骤，继续进行目标虚拟应用对象的确定；当这些待输出虚拟应用对象能够组成和牌的牌型时，可以认为待输出虚拟应用对象满足终止条件。
在一实施例中，如图8所示，该虚拟应用对象输出方法由于其准确性、以及灵活性，可以应用于“xx成河”、“xx到底”、“xx五星”等多种游 戏规则对应的虚拟应用，还可以应用于“单人游戏”、“双人游戏”、“三人游戏”、“四人游戏”等多种游戏形式，用户能够与其他用户组队进行游戏，提升游戏的趣味性。
由上可知，本申请实施例可以获取虚拟应用中多个虚拟应用对象的当前状态信息，当前状态信息用于指示虚拟应用对象处于已知状态或者处于未知状态，根据多个虚拟应用对象的当前状态信息，构建虚拟应用对象状态平面，其中，虚拟应用对象状态平面包括每个虚拟应用对象对应的区域，区域中包括其对应的虚拟应用对象的当前状态信息，根据虚拟应用对象状态平面，确定多个待输出虚拟应用对象各自对应的输出概率，根据多个待输出虚拟应用对象各自对应的输出概率，从多个待输出虚拟应用对象中确定目标虚拟应用对象进行输出。该方案通过将虚拟应用中各虚拟应用对象的当前状态信息用状态平面的方式进行表示，使得各虚拟应用对象的当前状态信息能够简明准确的表示在状态平面中，从而方便神经网络的识别与学习，以确定待输出虚拟对象对应的的输出概率。该方案还可以通过对待旋转状态子平面进行旋转的方式，使得整个状态平面更接近正方形，从而方便神经网络进行学习。再者，该方案通过在状态平面中插入隔离区域的方式，对需要进行区分的虚拟应用对象进行严格的区分，降低了番种误判的可能性，从而提高了目标虚拟应用对象输出的准确性。
根据前面实施例所描述的方法，以下将以该虚拟应用对象输出装置具体集成在网络设备举例作进一步详细说明。
参考图3，本申请实施例的虚拟应用对象输出方法的具体流程可以如下：
301、网络设备获取网络麻将游戏应用中多个虚拟麻将牌的当前状态信息。
在实际应用中，如图10所示，网络设备可以根据网络麻将游戏应用中的当前牌局，确定每个虚拟麻将牌此时处于已知状态还是未知状态，可以定义当前玩家自己已知的虚拟麻将牌、各个玩家已经输出的虚拟麻将牌、以及其他玩家展示出来的虚拟麻将牌等已知状态的虚拟麻将牌为已知状态。当虚拟麻将牌处于已知状态时，虚拟麻将牌对应的当前状态信息可以为“1”；可以定义其它位置状态的虚拟麻将牌为未知状态，当虚拟麻将牌处于未知状态时，虚拟麻将牌对应的当前状态信息可以为“0”。
302、网络设备根据多个虚拟麻将牌的当前状态信息，构建虚拟应用对象状态平面。
在实际应用中，网络设备可以根据每个虚拟麻将牌对应的当前状态信息，构建如图12所示的虚拟应用对象状态平面。该虚拟应用对象状态平面中包括多个矩形区域，每个矩形区域都代表一张虚拟麻将牌。多个矩形区域根据每个区域对应虚拟麻将牌的麻将牌名称，按照“一万、二万、三万、四 万、五万、六万、七万、八万、九万、一条、二条、三条、四条、五条、六条、七条、八条、九条、一筒、二筒、三筒、四筒、五筒、六筒、七筒、八筒、九筒、东、南、西、北、中、发、白”的顺序进行排列，整体排列为一个4*34的矩形阵列，其中矩形阵列中每行都包括34个名称不同的虚拟麻将牌，矩形阵列中每列都包括4个名称相同的虚拟麻将牌。每个矩形区域中都包括该矩形区域对应虚拟麻将牌的当前状态信息“0”或“1”。
303、网络设备将虚拟应用对象状态平面输入至概率获取网络中，获取多个待输出虚拟麻将牌各自对应的输出概率。
在实际应用中，如图10所示，获取到虚拟应用对象状态平面后，可以将该虚拟应用对象状态平面输入至CNN网络中，该CNN网络可以为概率获取网络，并获取得到当前玩家能够进行输出的每个待输出虚拟麻将牌对应的输出概率，如图7所示，当前玩家北风输出虚拟麻将牌“南”的概率为62.54％，当前玩家北风输出虚拟麻将牌“中”的概率为25％，当前玩家北风输出虚拟麻将牌“北”的概率为7.76％，当前玩家北风输出虚拟麻将牌“西”的概率为4.34％，当前玩家北风输出虚拟麻将牌“东”的概率为0.21％，当前玩家北风输出虚拟麻将牌“幺鸡”的概率为0.13％，等等。
304、网络设备根据多个待输出虚拟麻将牌各自对应的输出概率，从多个待输出虚拟麻将牌中确定目标虚拟麻将牌进行输出。
在实际应用中，获取到每个待输出虚拟麻将牌对应的输出概率后，可以通过对概率分布进行采样，从多个待输出虚拟麻将牌中确定出目标虚拟麻将牌，并将目标虚拟麻将牌进行输出。
305、网络设备更新待输出虚拟麻将牌。
在实际应用中，网络设备输出目标虚拟麻将牌后，待输出虚拟麻将牌发生了变化，此时，可以对待输出虚拟麻将牌进行更新，以便对更新后的待输出虚拟麻将牌进行分析。
306、当待输出虚拟麻将牌不满足终止条件时，网络设备更新虚拟麻将牌的当前状态信息。
在实际应用中，在当前玩家能够支配的待输出虚拟麻将牌没有组成和牌牌型时，游戏继续，网络设备可以根据当前的牌局状态，对当前牌局中每个虚拟麻将牌的当前状态信息进行更新。
307、网络设备返回执行根据多个虚拟应用对象的当前状态信息，构建虚拟应用对象状态平面的步骤。
在实际应用中，当每个虚拟麻将牌的当前状态信息都更新完毕后，可以返回根据多个虚拟应用对象的当前状态信息，构建虚拟应用对象状态平面的步骤，继续进行目标虚拟应用对象的识别，直至当前玩家能够支配的待输出虚拟麻将牌已经组成了和牌牌型，此时，游戏流程可以终止。
由上可知，本申请实施例可以通过网络设备获取网络麻将游戏应用中多 个虚拟麻将牌的当前状态信息，根据多个虚拟麻将牌的当前状态信息，构建虚拟应用对象状态平面，将虚拟应用对象状态平面输入至概率获取网络中，获取多个待输出虚拟麻将牌各自对应的输出概率，根据多个待输出虚拟麻将牌各自对应的输出概率，从多个待输出虚拟麻将牌中确定目标虚拟麻将牌进行输出，更新待输出虚拟麻将牌，当待输出虚拟麻将牌不满足终止条件时，更新虚拟麻将牌的当前状态信息，返回执行根据多个虚拟麻将牌的当前状态信息，构建虚拟应用对象状态平面的步骤。该方案通过将虚拟应用中的虚拟应用对象的当前状态信息用状态平面的方式进行表示，使得虚拟应用对象的当前状态信息能够简明准确的表示在状态平面中，从而方便神经网络的识别与学习，确定待输出虚拟应用对象对应的输出概率，进而提高了目标虚拟应用对象输出的准确性。
根据前面实施例所描述的方法，以下将以该虚拟应用对象输出装置具体集成在网络设备举例作进一步详细说明。
参考图4，本申请实施例的虚拟应用对象输出方法的具体流程可以如下：
401、网络设备获取网络麻将游戏应用中多个虚拟麻将牌的当前状态信息、以及多个虚拟麻将牌的对象类型。
在实际应用中，网络设备可以获取网络麻将游戏应用中每个虚拟麻将牌的当前状态信息“0”或“1”，并且根据每个虚拟麻将牌的麻将牌名称，将虚拟麻将牌分为“万”、“条”、“筒”、“风”、以及“箭”五种对象类型，然后确定每个虚拟麻将牌对应的对象类型。
402、网络设备根据多个虚拟麻将牌的当前状态信息、以及对象类型，构建虚拟应用对象状态平面。
在实际应用中，网络设备可以根据每个虚拟麻将牌对应的当前状态信息、以及对象类型，构建如图15所示的虚拟应用对象状态平面。该虚拟应用对象状态平面为4*38的矩形阵列，从左向右数第10列、第20列、第30列、以及第35列的区域都为隔离区域。从左向右数第1列至第9列的状态子平面对应对象类型“万”，第11列至第19列的状态子平面对应对象类型“条”，第21列至第29列的状态子平面对应对象类型“筒”，第31列至第34列的状态子平面对应对象类型“风”，第36列至第38列的状态子平面对应对象类型“箭”。
403、网络设备将虚拟应用对象状态平面输入至概率获取网络中，获取多个待输出虚拟麻将牌各自对应的输出概率。
404、网络设备根据多个待输出虚拟麻将牌各自对应的输出概率，从多个待输出虚拟麻将牌中确定目标虚拟麻将牌进行输出。
405、网络设备更新待输出虚拟麻将牌。
406、当待输出虚拟麻将牌不满足终止条件时，网络设备更新虚拟麻将 牌的当前状态信息。
407、网络设备返回执行根据当前状态信息，构建虚拟应用对象状态平面的步骤。
上述根据神经网络确定目标虚拟应用对象的步骤在前文中已经叙述，此处不再赘述。
由上可知，本申请实施例可以通过网络设备获取网络麻将游戏应用中多个虚拟麻将牌的当前状态信息、以及多个虚拟麻将牌的对象类型，根据当前状态信息、以及对象类型，构建虚拟应用对象状态平面，将虚拟应用对象状态平面输入至概率获取网络中，获取待输出虚拟麻将牌的输出概率，根据待输出虚拟麻将牌的输出概率，从多个待输出虚拟麻将牌中确定目标虚拟麻将牌进行输出，更新待输出虚拟麻将牌，当待输出虚拟麻将牌不满足终止条件时，更新虚拟麻将牌的当前状态信息，返回执行根据当前状态信息，构建虚拟应用对象状态平面的步骤。该方案通过将虚拟应用中的虚拟应用对象的当前状态信息用状态平面的方式进行表示，使得当前状态信息能够简明准确的表示在状态平面中，从而方便神经网络的识别与学习，确定待输出虚拟应用对象对应的输出概率，进而提高了目标虚拟应用对象输出的准确性。
参考图5，本申请实施例的虚拟应用对象输出方法的具体流程可以如下：
501、网络设备获取网络麻将游戏应用中多个虚拟麻将牌的当前状态信息、多个虚拟麻将牌的对象类型、需要进行隔离的目标对象子类型、以及两个目标对象子类型之间隔离区域的区域尺寸。
在实际应用中，网络设备可以确定五个对象类型“万”、“条”、“筒”、“风”、以及“箭”中，需要进行隔离的目标对象类型“风”和“箭”，然后将对象类型“风”划分为“东”、“南”、“西”、以及“北”四种目标对象子类型，将对象类型“箭”划分为“中”、“发”、以及“白”三种目标对象子类型。并确定目标对象子类型“东”与“南”之间隔离区域的区域尺寸为1，目标对象子类型“北”与“中”之间隔离区域的区域尺寸为2，目标对象子类型“中”与“发”之间隔离区域的区域尺寸为2，等等。
502、网络设备根据当前状态信息、对象类型、目标对象子类型、以及区域尺寸，构建虚拟应用对象状态平面。
在实际应用中，网络设备可以根据每个虚拟麻将牌对应的当前状态信息、对象类型、目标对象子类型、以及区域尺寸，构建如图18所示的虚拟应用对象状态平面。该虚拟应用对象状态平面为4*46的矩形阵列，从左向右数第10列、第20列、第30列、第32列、第34列、第36列、第38列、第39列、第41列、第42列、第44列、第45列的区域都为隔离区域。从左向右数第1列至第9列的状态子平面对应对象类型“万”，第11列至第19列的状态子平面对应对象类型“条”，第21列至第29列的状态子平面 对应对象类型“筒”，第31列的隔离后状态区域对应目标对象子类型“东”，第33列的隔离后状态区域对应目标对象子类型“南”，第35列的隔离后状态区域对应目标对象子类型“西”，第37列的隔离后状态区域对应目标对象子类型“北”，第40列的隔离后状态区域对应目标对象子类型“中”，第43列的隔离后状态区域对应目标对象子类型“发”，第46列的隔离后状态区域对应目标对象子类型“白”。
503、网络设备将虚拟应用对象状态平面输入至概率获取网络中，获取待输出虚拟麻将牌的输出概率。
504、网络设备根据待输出虚拟麻将牌的输出概率，从多个待输出虚拟麻将牌中确定目标虚拟麻将牌进行输出。
505、网络设备更新待输出虚拟麻将牌。
506、当待输出虚拟麻将牌不满足终止条件时，网络设备更新虚拟麻将牌的当前状态信息。
507、网络设备返回执行根据当前状态信息，构建虚拟应用对象状态平面的步骤。
上述根据神经网络确定目标虚拟应用对象的步骤在前文中已经叙述，此处不再赘述。
由上可知，本申请实施例可以通过网络设备获取网络麻将游戏应用中多个虚拟麻将牌的当前状态信息、多个虚拟麻将牌的对象类型、需要进行隔离的目标对象子类型、以及两个目标对象子类型之间隔离区域的区域尺寸，根据当前状态信息、对象类型、目标对象子类型、以及区域尺寸，构建虚拟应用对象状态平面，将虚拟应用对象状态平面输入至概率获取网络中，获取待输出虚拟麻将牌的输出概率，根据待输出虚拟麻将牌的输出概率，从多个待输出虚拟麻将牌中确定目标虚拟麻将牌进行输出，更新待输出虚拟麻将牌，当待输出虚拟麻将牌不满足终止条件时，更新虚拟麻将牌的当前状态信息，返回执行根据当前状态信息，构建虚拟应用对象状态平面的步骤。该方案通过将虚拟应用中的虚拟应用对象的当前状态信息用状态平面的方式进行表示，使得当前状态信息能够简明准确的表示在状态平面中，从而方便神经网络的识别与学习，确定待输出虚拟应用对象对应的输出概率。该方案通过在状态平面中插入隔离区域的方式，对需要进行区分的虚拟应用对象进行严格的区分，降低了番种误判的可能性，从而提高了目标虚拟应用对象输出的准确性。
参考图6，本申请实施例的虚拟应用对象输出方法的具体流程可以如下：
601、网络设备获取网络麻将游戏应用中多个虚拟麻将牌的当前状态信息、多个虚拟麻将牌的对象类型、需要进行隔离的目标对象子类型、以及两个目标对象子类型之间隔离区域的区域尺寸。
602、网络设备根据当前状态信息、对象类型、目标对象子类型、以及区域尺寸，构建虚拟应用对象状态平面。
在实际应用中，网络设备可以根据每个虚拟麻将牌对应的当前状态信息、对象类型、目标对象子类型、以及区域尺寸，构建如图21a所示的虚拟应用对象状态平面。该虚拟应用对象状态平面中包括5个旋转后状态子平面、以及多个隔离区域，每个旋转后状态子平面对应一种对象类型，隔离区域将不同的对象类型、以及不同的目标对象子类型隔离开。图中虚拟应用对象状态平面为9*24的矩形阵列，纵向方向表示虚拟麻将牌的牌面值1至9，横向方向分别为对应“万”、“条”、“筒”、“风”、以及“箭”的5个旋转后状态子平面。
603、网络设备将虚拟应用对象状态平面输入至概率获取网络中，获取待输出虚拟麻将牌的输出概率。
604、网络设备根据待输出虚拟麻将牌的输出概率，从多个待输出虚拟麻将牌中确定目标虚拟麻将牌进行输出。
605、网络设备更新待输出虚拟麻将牌。
606、当待输出虚拟麻将牌不满足终止条件时，网络设备更新虚拟麻将牌的当前状态信息。
607、网络设备返回执行根据当前状态信息，构建虚拟应用对象状态平面的步骤。
上述根据神经网络确定目标虚拟应用对象的步骤在前文中已经叙述，此处不再赘述。
由上可知，本申请实施例可以通过网络设备获取网络麻将游戏应用中多个虚拟麻将牌的当前状态信息、多个虚拟麻将牌的对象类型、需要进行隔离的目标对象子类型、以及两个目标对象子类型之间隔离区域的区域尺寸，根据当前状态信息、对象类型、目标对象子类型、以及区域尺寸，构建虚拟应用对象状态平面，将虚拟应用对象状态平面输入至概率获取网络中，获取待输出虚拟麻将牌的输出概率，根据待输出虚拟麻将牌的输出概率，从多个待输出虚拟麻将牌中确定目标虚拟麻将牌进行输出，更新待输出虚拟麻将牌，当待输出虚拟麻将牌不满足终止条件时，更新虚拟麻将牌的当前状态信息，返回执行根据当前状态信息，构建虚拟应用对象状态平面的步骤。该方案通过将虚拟应用中的虚拟应用对象的当前状态信息用状态平面的方式进行表示，使得当前状态信息能够简明准确的表示在状态平面中，从而方便神经网络的识别与学习，确定待输出虚拟应用对象对应的输出概率。该方案还可以通过对待旋转状态子平面进行旋转的方式，使得整个状态平面更接近正方形，从而方便神经网络进行学习。再者，该方案通过在状态平面中插入隔离区域的方式，对需要进行区分的虚拟应用对象进行严格的区分，降低了番种误判的可能性，从而提高了目标虚拟应用对象输出的准确性。
为了更好地实施以上方法，本申请实施例还可以提供一种虚拟应用对象输出装置，该虚拟应用对象输出装置具体可以集成在网络设备中，该网络设备可以包括服务器、终端等，其中，终端可以包括：手机、平板电脑、笔记本电脑或个人计算机(PC，Personal Computer)等。
例如，如图22所示，该虚拟应用对象输出装置可以包括获取模块221、构建模块222、概率确定模块223和输出模块224，如下：
获取模块221，用于获取虚拟应用中多个虚拟应用对象的当前状态信息，所述当前状态信息用于指示虚拟应用对象处于已知状态或者处于未知状态；
构建模块222，用于根据所述多个虚拟应用对象的当前状态信息，构建虚拟应用对象状态平面，其中，所述虚拟应用对象状态平面包括每个虚拟应用对象对应的区域，所述区域中包括其对应的所述虚拟应用对象的当前状态信息；
概率确定模块223，用于根据所述虚拟应用对象状态平面，确定多个待输出虚拟应用对象各自对应的输出概率；
输出模块224，用于根据所述多个待输出虚拟应用对象各自对应的输出概率，从所述多个待输出虚拟应用对象中确定目标虚拟应用对象进行输出。
在一实施例中，所述构建模块222可以包括第一获取子模块2221和第一构建子模块2222，如下：
第一获取子模块2221，用于获取多个虚拟应用对象的对象类型；
第一构建子模块2222，用于根据所述当前状态信息、以及所述对象类型，构建虚拟应用对象状态平面，其中，所述虚拟应用对象状态平面包括若干状态子平面、以及若干隔离区域，所述状态子平面与所述对象类型相对应，所述隔离区域位于相邻的两个状态子平面之间。
在一实施例中，所述第一构建子模块2222，可以具体用于：
根据所述当前状态信息，构建第一初始状态平面；
根据所述对象类型，将所述第一初始状态平面分割为若干状态子平面；
在两个相邻的状态子平面之间插入隔离区域，得到虚拟应用对象状态平面。
在一实施例中，所述第一构建子模块2222可以包括第一确定子模块22221和第二构建子模块22222，如下：
第一确定子模块22221，用于从若干对象类型中，确定需要进行隔离的目标对象类型，所述目标对象类型包括多个目标对象子类型；
第二构建子模块22222，用于根据所述当前状态信息、所述对象类型、以及所述目标对象子类型，构建虚拟应用对象状态平面，其中，所述虚拟应用对象状态平面中的状态子平面包括若干隔离后状态区域，所述隔离后状态区域与所述目标对象子类型相对应，所述虚拟应用对象状态平面中的隔离区 域还位于相邻的两个隔离后状态区域之间。
在一实施例中，所述第二构建子模块22222可以包括第二确定子模块222221和第三构建子模块222222，如下：
第二确定子模块222221，用于确定需要进行隔离的两个目标对象子类型之间的隔离区域的区域尺寸；
第三构建子模块222222，用于根据所述当前状态信息、所述对象类型、所述目标对象子类型、以及所述区域尺寸，构建虚拟应用对象状态平面，其中，所述虚拟应用对象状态平面中的两个隔离后状态区域之间包括区域尺寸的隔离区域。
在一实施例中，所述第三构建子模块222222，可以具体用于：
根据所述当前状态信息、以及所述对象类型，构建第二初始状态平面；
确定所述第二初始状态平面中所述目标对象类型对应的目标状态子平面；
根据所述目标对象子类型，将所述目标状态子平面分割为若干隔离后状态区域；
在两个相邻的隔离后状态区域之间插入所述区域尺寸的隔离区域，得到隔离后初始状态平面；
根据所述隔离后初始状态平面、以及所述第二初始状态平面，构建虚拟应用对象状态平面。
在一实施例中，所述第三构建子模块222222，可以具体用于：
根据所述当前状态信息、所述对象类型、所述目标对象子类型、以及所述区域尺寸，构建第三初始状态平面；
根据所述对象类型，将所述第三初始状态平面分割为若干待旋转状态子平面；
将每个待旋转状态子平面进行旋转，得到旋转后状态子平面；
在相邻两个旋转后状态子平面之间插入隔离区域，得到虚拟应用对象状态平面。
在一实施例中，所述虚拟应用对象输出装置还可以包括第一更新模块2251、第二更新模块2252、返回模块2253和终止模块2254，如下：
第一更新模块2251，用于更新待输出虚拟应用对象；
第二更新模块2252，用于当所述待输出虚拟应用对象不满足终止条件时，更新虚拟应用对象的当前状态信息；
返回模块2253，用于返回执行根据所述多个虚拟应用对象的当前状态信息，构建虚拟应用对象状态平面的步骤；
终止模块2254，用于当所述待输出虚拟应用对象满足终止条件时，终止输出。
在一实施例中，所述概率获取模块223，可以具体用于：
将所述虚拟应用对象状态平面输入至概率获取网络；
基于所述概率获取网络，获取待输出虚拟应用对象对应的输出概率。
在一实施例中，所述虚拟应用对象输出装置还可以包括第二获取子模块2261、训练子模块2262和收敛子模块2263，如下：
第二获取子模块2261，用于获取若干样本状态平面、以及每个样本状态平面对应的真实输出对象；
训练子模块2262，用于根据所述样本状态平面对预设概率获取网络进行训练，得到所述样本状态平面对应的预测输出对象；
收敛子模块2263，用于根据所述样本状态平面对应的真实输出对象和所述预测输出对象，对所述预设概率获取网络进行收敛，得到概率获取网络。
具体实施时，以上各个单元可以作为独立的实体来实现，也可以进行任意组合，作为同一或若干个实体来实现，以上各个单元的具体实施可参见前面的方法实施例，在此不再赘述。
由上可知，本实施例的虚拟应用对象输出装置通过获取模块221获取虚拟应用中多个虚拟应用对象的当前状态信息，当前状态信息用于指示虚拟应用对象处于已知状态或者处于未知状态，通过构建模块222根据多个虚拟应用对象的当前状态信息，构建虚拟应用对象状态平面，其中，虚拟应用对象状态平面包括每个虚拟应用对象对应的区域，区域中包括其对应的虚拟应用对象的当前状态信息，通过概率确定模块223根据虚拟应用对象状态平面，确定多个待输出虚拟应用对象各自对应的输出概率，通过输出模块224根据多个待输出虚拟应用对象各自对应的输出概率，从多个待输出虚拟应用对象中确定目标虚拟应用对象进行输出。该方案通过将虚拟应用中的虚拟应用对象的当前状态信息用状态平面的方式进行表示，使得当前状态信息能够简明准确的表示在状态平面中，从而方便神经网络的识别与学习，确定多个待输出虚拟应用对象各自对应的输出概率。该方案还可以通过对待旋转状态子平面进行旋转的方式，使得整个状态平面更接近正方形，从而方便神经网络进行学习。再者，该方案通过在状态平面中插入隔离区域的方式，对需要进行区分的虚拟应用对象进行严格的区分，降低了番种误判的可能性，从而提高了目标虚拟应用对象输出的准确性。
本申请实施例还提供一种网络设备，该网络设备可以集成本申请实施例所提供的任一种虚拟应用对象输出装置。
例如，如图23所示，其示出了本申请实施例所涉及的网络设备的结构示意图，具体来讲：
该网络设备可以包括一个或者一个以上处理核心的处理器231、一个或一个以上计算机可读存储介质的存储器232、电源233和输入单元234等部件。本领域技术人员可以理解，图23中示出的网络设备结构并不构成对网络设备的限定，可以包括比图示更多或更少的部件，或者组合某些部件，或 者不同的部件布置。其中：
处理器231是该网络设备的控制中心，利用各种接口和线路连接整个网络设备的各个部分，通过运行或执行存储在存储器232内的软件程序和/或模块，以及调用存储在存储器232内的数据，执行网络设备的各种功能和处理数据，从而对网络设备进行整体监控。可选的，处理器231可包括一个或多个处理核心；优选的，处理器231可集成应用处理器和调制解调处理器，其中，应用处理器主要处理操作系统、用户界面和应用程序等，调制解调处理器主要处理无线通信。可以理解的是，上述调制解调处理器也可以不集成到处理器231中。
存储器232可用于存储软件程序以及模块，处理器231通过运行存储在存储器232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。存储器232可主要包括存储程序区和存储数据区，其中，存储程序区可存储操作系统、至少一个功能所需的应用程序(比如声音播放功能、图像播放功能等)等；存储数据区可存储根据网络设备的使用所创建的数据等。此外，存储器232可以包括高速随机存取存储器，还可以包括非易失性存储器，例如至少一个磁盘存储器件、闪存器件、或其他易失性固态存储器件。相应地，存储器232还可以包括存储器控制器，以提供处理器231对存储器232的访问。
网络设备还包括给各个部件供电的电源233，优选的，电源233可以通过电源管理系统与处理器231逻辑相连，从而通过电源管理系统实现管理充电、放电、以及功耗管理等功能。电源233还可以包括一个或一个以上的直流或交流电源、再充电系统、电源故障检测电路、电源转换器或者逆变器、电源状态指示器等任意组件。
该网络设备还可包括输入单元234，该输入单元234可用于接收输入的数字或字符信息，以及产生与用户设置以及功能控制有关的键盘、鼠标、操作杆、光学或者轨迹球信号输入。
尽管未示出，网络设备还可以包括显示单元等，在此不再赘述。具体在本实施例中，网络设备中的处理器231会按照如下的指令，将一个或一个以上的应用程序的进程对应的可执行文件加载到存储器232中，并由处理器231来运行存储在存储器232中的应用程序，从而实现各种功能，如下：
获取虚拟应用中多个虚拟应用对象的当前状态信息，当前状态信息用于指示虚拟应用对象处于已知状态或者处于未知状态，根据多个虚拟应用对象的当前状态信息，构建虚拟应用对象状态平面，其中，虚拟应用对象状态平面包括每个虚拟应用对象对应的区域，区域中包括其对应的虚拟应用对象的当前状态信息，根据虚拟应用对象状态平面，确定多个待输出虚拟应用对象各自对应的输出概率，根据多个待输出虚拟应用对象各自对应的输出概率，从多个待输出虚拟应用对象中确定目标虚拟应用对象进行输出。
以上各个操作的具体实施可参见前面的实施例，在此不再赘述。
由上可知，本申请实施例的网络设备可以获取虚拟应用中多个虚拟应用对象的当前状态信息，当前状态信息用于指示虚拟应用对象处于已知状态或者处于未知状态，根据多个虚拟应用对象的当前状态信息，构建虚拟应用对象状态平面，其中，虚拟应用对象状态平面包括每个虚拟应用对象对应的区域，区域中包括其对应的虚拟应用对象的当前状态信息，根据虚拟应用对象状态平面，确定多个待输出虚拟应用对象各自对应的输出概率，根据多个待输出虚拟应用对象各自对应的输出概率，从多个待输出虚拟应用对象中确定目标虚拟应用对象进行输出。该方案通过将虚拟应用中的虚拟应用对象的当前状态信息用状态平面的方式进行表示，使得当前状态信息能够简明准确的表示在状态平面中，从而方便神经网络的识别与学习，以确定待输出虚拟应用对象对应的输出概率。该方案还可以通过对待旋转状态子平面进行旋转的方式，使得整个状态平面更接近正方形，从而方便神经网络进行学习。再者，该方案通过在状态平面中插入隔离区域的方式，对需要进行区分的虚拟应用对象进行严格的区分，降低了番种误判的可能性，从而提高了目标虚拟应用对象输出的准确性。
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可以通过指令来完成，或通过指令控制相关的硬件来完成，该指令可以存储于一计算机可读存储介质中，并由处理器进行加载和执行。
为此，本申请实施例提供一种存储介质，其中存储有多条指令，该指令能够被处理器进行加载，以执行本申请实施例所提供的任一种虚拟应用对象输出方法中的步骤。例如，该指令可以执行如下步骤：
获取虚拟应用中多个虚拟应用对象的当前状态信息，当前状态信息用于指示虚拟应用对象处于已知状态或者处于未知状态，根据多个虚拟应用对象的当前状态信息，构建虚拟应用对象状态平面，其中，虚拟应用对象状态平面包括每个虚拟应用对象对应的区域，区域中包括其对应的虚拟应用对象的当前状态信息，根据虚拟应用对象状态平面，确定多个待输出虚拟应用对象各自对应的输出概率，根据待输出虚拟应用对象各自对应的输出概率，从多个待输出虚拟应用对象中确定目标虚拟应用对象进行输出。
以上各个操作的具体实施可参见前面的实施例，在此不再赘述。
其中，该存储介质可以包括：只读存储器(ROM，Read Only Memory)、随机存取记忆体(RAM，Random Access Memory)、磁盘或光盘等。
由于该存储介质中所存储的指令，可以执行本申请实施例所提供的任一种虚拟应用对象输出方法中的步骤，因此，可以实现本申请实施例所提供的任一种虚拟应用对象输出方法所能实现的有益效果，详见前面的实施例，在此不再赘述。
以上对本申请实施例所提供的一种虚拟应用对象输出方法、装置以及计 算机存储介质进行了详细介绍，本文中应用了具体个例对本申请的原理及实施方式进行了阐述，以上实施例的说明只是用于帮助理解本申请的方法及其核心思想；同时，对于本领域的技术人员，依据本申请的思想，在具体实施方式及应用范围上均会有改变之处，综上所述，本说明书内容不应理解为对本申请的限制。

Claims (13)

  1. 一种虚拟应用对象输出方法，由网络设备执行，包括：
    获取虚拟应用中多个虚拟应用对象的当前状态信息，所述当前状态信息用于指示虚拟应用对象处于已知状态或者处于未知状态；
    根据所述多个虚拟应用对象的当前状态信息，构建虚拟应用对象状态平面，其中，所述虚拟应用对象状态平面包括每个虚拟应用对象对应的区域，所述区域中包括其对应的所述虚拟应用对象的当前状态信息；
    根据所述虚拟应用对象状态平面，确定多个待输出虚拟应用对象各自对应的输出概率；
    根据所述多个待输出虚拟应用对象各自对应的输出概率，从所述多个待输出虚拟应用对象中确定目标虚拟应用对象进行输出。
  2. 根据权利要求1所述的虚拟应用对象输出方法，所述根据所述多个虚拟应用对象的当前状态信息，构建虚拟应用对象状态平面，包括：
    获取所述多个虚拟应用对象的对象类型；
    根据所述当前状态信息、以及所述对象类型，构建所述虚拟应用对象状态平面，其中，所述虚拟应用对象状态平面包括若干状态子平面、以及若干隔离区域，所述状态子平面与所述对象类型相对应，所述隔离区域位于相邻的两个所述状态子平面之间。
  3. 根据权利要求2所述的虚拟应用对象输出方法，所述根据所述当前状态信息、以及所述对象类型，构建所述虚拟应用对象状态平面，包括：
    根据所述当前状态信息，构建第一初始状态平面；
    根据所述对象类型，将所述第一初始状态平面分割为所述若干状态子平面；
    在两个相邻的所述状态子平面之间插入所述隔离区域，得到所述虚拟应用对象状态平面。
  4. 根据权利要求2所述的虚拟应用对象输出方法，所述根据所述当前状态信息、以及所述对象类型，构建所述虚拟应用对象状态平面，包括：
    从若干对象类型中，确定需要进行隔离的目标对象类型，所述目标对象类型包括多个目标对象子类型；
    根据所述当前状态信息、所述对象类型、以及所述目标对象子类型，构建所述虚拟应用对象状态平面，其中，所述虚拟应用对象状态平面中的状态子平面包括若干隔离后状态区域，所述隔离后状态区域与所述目标对象子类型相对应，所述虚拟应用对象状态平面中的隔离区域还位于相邻的两个所述隔离后状态区域之间。
  5. 根据权利要求4所述的虚拟应用对象输出方法，所述根据所述当前状态信息、所述对象类型、以及所述目标对象子类型，构建所述虚拟应用对象状态平面，包括：
    确定需要进行隔离的两个目标对象子类型之间的隔离区域的区域尺寸；
    根据所述当前状态信息、所述对象类型、所述目标对象子类型、以及所述区域尺寸，构建所述虚拟应用对象状态平面，其中，所述虚拟应用对象状态平面中的两个所述隔离后状态区域之间包括所述区域尺寸的隔离区域。
  6. 根据权利要求5所述的虚拟应用对象输出方法，所述根据所述当前状态信息、所述对象类型、所述目标对象子类型、以及所述区域尺寸，构建所述虚拟应用对象状态平面，包括：
    根据所述当前状态信息、以及所述对象类型，构建第二初始状态平面；
    确定所述第二初始状态平面中所述目标对象类型对应的目标状态子平面；
    根据所述目标对象子类型，将所述目标状态子平面分割为若干隔离后状态区域；
    在两个相邻的所述隔离后状态区域之间插入所述区域尺寸的隔离区域，得到隔离后初始状态平面；
    根据所述隔离后初始状态平面、以及所述第二初始状态平面，构建所述虚拟应用对象状态平面。
  7. 根据权利要求5所述的虚拟应用对象输出方法，所述根据所述当前状态信息、所述对象类型、所述目标对象子类型、以及所述区域尺寸，构建所述虚拟应用对象状态平面，包括：
    根据所述当前状态信息、所述对象类型、所述目标对象子类型、以及所述区域尺寸，构建第三初始状态平面；
    根据所述对象类型，将所述第三初始状态平面分割为若干待旋转状态子平面；
    将每个所述待旋转状态子平面进行旋转，得到旋转后状态子平面；
    在相邻两个所述旋转后状态子平面之间插入隔离区域，得到所述虚拟应用对象状态平面。
  8. 根据权利要求1所述的虚拟应用对象输出方法，在所述根据所述多个待输出虚拟应用对象各自对应的输出概率，从所述多个待输出虚拟应用对象中确定目标虚拟应用对象进行输出之后，还包括：
    更新所述待输出虚拟应用对象；
    当所述待输出虚拟应用对象不满足终止条件时，更新所述虚拟应用对象的当前状态信息；返回执行所述根据所述多个虚拟应用对象的当前状态信息，构建虚拟应用对象状态平面的步骤；
    当所述待输出虚拟应用对象满足终止条件时，终止输出。
  9. 根据权利要求1所述的虚拟应用对象输出方法，所述根据所述虚拟应用对象状态平面，确定所述多个待输出虚拟应用对象各自对应的输出概率，包括：
    将所述虚拟应用对象状态平面输入至概率获取网络；
    基于所述概率获取网络，获取所述待输出虚拟应用对象对应的输出概率。
  10. 根据权利要求9所述的虚拟应用对象输出方法，所述方法还包括：
    获取若干样本状态平面、以及每个样本状态平面对应的真实输出对象；
    根据所述样本状态平面对预设概率获取网络进行训练，得到所述样本状态平面对应的预测输出对象；
    根据所述样本状态平面对应的真实输出对象和所述预测输出对象，对所述预设概率获取网络进行收敛，得到所述概率获取网络。
  11. 一种虚拟应用对象输出装置，包括：
    获取模块，用于获取虚拟应用中多个虚拟应用对象的当前状态信息，所述当前状态信息用于指示虚拟应用对象处于已知状态或者处于未知状态；
    构建模块，用于根据所述多个虚拟应用对象的当前状态信息，构建虚拟应用对象状态平面，其中，所述虚拟应用对象状态平面包括每个虚拟应用对象对应的区域，所述区域中包括其对应的所述虚拟应用对象的当前状态信息；
    概率确定模块，用于根据所述虚拟应用对象状态平面，确定多个待输出虚拟应用对象各自对应的输出概率；
    输出模块，用于根据所述多个待输出虚拟应用对象各自对应的输出概率，从所述多个待输出虚拟应用对象中确定目标虚拟应用对象进行输出。
  12. 一种计算机存储介质，其上存储有计算机程序，当所述计算机程序在计算机上运行时，使得所述计算机执行如权利要求1-10任一项所述的虚拟应用对象输出方法。
  13. 一种计算机程序产品，包括指令，当其在计算机上运行时，使得计算机执行如权利要求1-10任一项所述的虚拟应用对象输出方法。
PCT/CN2020/115219 2019-09-26 2020-09-15 一种虚拟应用对象输出方法、装置以及计算机存储介质 WO2021057539A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
JP2021563356A JP7408213B2 (ja) 2019-09-26 2020-09-15 仮想アプリケーションオブジェクトの出力方法、装置及びコンピュータプログラム
US17/460,610 US11704980B2 (en) 2019-09-26 2021-08-30 Method, apparatus, and computer storage medium for outputting virtual application object

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910914842.0A CN110721471B (zh) 2019-09-26 2019-09-26 一种虚拟应用对象输出方法、装置以及计算机存储介质
CN201910914842.0 2019-09-26

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US17/460,610 Continuation US11704980B2 (en) 2019-09-26 2021-08-30 Method, apparatus, and computer storage medium for outputting virtual application object

Publications (1)

Publication Number Publication Date
WO2021057539A1 true WO2021057539A1 (zh) 2021-04-01

Family

ID=69219469

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/115219 WO2021057539A1 (zh) 2019-09-26 2020-09-15 一种虚拟应用对象输出方法、装置以及计算机存储介质

Country Status (4)

Country Link
US (1) US11704980B2 (zh)
JP (1) JP7408213B2 (zh)
CN (1) CN110721471B (zh)
WO (1) WO2021057539A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110721471B (zh) 2019-09-26 2023-10-31 腾讯科技（深圳）有限公司 一种虚拟应用对象输出方法、装置以及计算机存储介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103810379A (zh) * 2014-01-23 2014-05-21 珠海多玩信息技术有限公司 目标状态推荐方法及装置
CN105653831A (zh) * 2014-11-10 2016-06-08 博雅网络游戏开发（深圳）有限公司 电子牌游戏牌型推荐方法和装置
CN107648853A (zh) * 2017-08-15 2018-02-02 腾讯科技（深圳）有限公司 在游戏界面中显示目标对象方法、装置以及存储介质
CN110598182A (zh) * 2019-09-11 2019-12-20 腾讯科技（深圳）有限公司 一种信息预测的方法及相关设备
CN110721471A (zh) * 2019-09-26 2020-01-24 腾讯科技（深圳）有限公司 一种虚拟应用对象输出方法、装置以及计算机存储介质

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7425176B2 (en) * 2005-09-10 2008-09-16 Bally Gaming, Inc. Simulated poker with bonus wheel adder
CN104166779B (zh) * 2013-05-17 2018-09-07 腾讯科技（深圳）有限公司 在游戏终端中模拟纸牌游戏的实现方法及装置
CN107908350A (zh) * 2017-12-28 2018-04-13 努比亚技术有限公司 一种解锁方法、终端及计算机可读存储介质
CN108888958B (zh) * 2018-06-22 2023-03-21 深圳市腾讯网络信息技术有限公司 虚拟场景中的虚拟对象控制方法、装置、设备及存储介质
CN109760067B (zh) * 2019-01-22 2021-05-07 中国科学院大学 会打纸牌的智能机器人系统以及设备
US11335058B2 (en) * 2020-10-13 2022-05-17 Electronic Arts Inc. Spatial partitioning for graphics rendering

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103810379A (zh) * 2014-01-23 2014-05-21 珠海多玩信息技术有限公司 目标状态推荐方法及装置
CN105653831A (zh) * 2014-11-10 2016-06-08 博雅网络游戏开发（深圳）有限公司 电子牌游戏牌型推荐方法和装置
CN107648853A (zh) * 2017-08-15 2018-02-02 腾讯科技（深圳）有限公司 在游戏界面中显示目标对象方法、装置以及存储介质
CN110598182A (zh) * 2019-09-11 2019-12-20 腾讯科技（深圳）有限公司 一种信息预测的方法及相关设备
CN110721471A (zh) * 2019-09-26 2020-01-24 腾讯科技（深圳）有限公司 一种虚拟应用对象输出方法、装置以及计算机存储介质

Also Published As

Publication number Publication date
US20210390833A1 (en) 2021-12-16
JP7408213B2 (ja) 2024-01-05
US11704980B2 (en) 2023-07-18
CN110721471A (zh) 2020-01-24
JP2022530884A (ja) 2022-07-04
CN110721471B (zh) 2023-10-31

Similar Documents

Publication Publication Date Title
CN109621422B (zh) 电子棋牌决策模型训练方法及装置、策略生成方法及装置
JP7253570B2 (ja) 遠隔ユーザ入力に基づく文脈インゲーム要素認識、注釈及び対話
US11938403B2 (en) Game character behavior control method and apparatus, storage medium, and electronic device
KR101430930B1 (ko) 클라우드 스트리밍 기반의 게임 제공 방법, 시스템, 클라이언트 단말기 및 서비스장치
CN107066239A (zh) 一种实现卷积神经网络前向计算的硬件结构
JP2023162233A (ja) 仮想オブジェクトの制御方法、装置、端末及び記憶媒体
CN113599798B (zh) 基于深度强化学习方法的中国象棋博弈学习方法及系统
CN111405360B (zh) 视频处理方法、装置、电子设备和存储介质
DE112021000225T5 (de) Auflösungshochskalierung zur Ereigniserkennung
CN109284812A (zh) 一种基于改进dqn的视频游戏模拟方法
WO2020233709A1 (zh) 模型压缩方法及装置
WO2023273813A1 (zh) 一种交互方法、装置、计算机设备以及存储介质
JP7085600B2 (ja) 画像間の類似度を利用した類似領域強調方法およびシステム
CN107952241A (zh) 渲染控制方法、装置及可读存储介质
WO2021057539A1 (zh) 一种虚拟应用对象输出方法、装置以及计算机存储介质
CN111672117A (zh) 虚拟对象的选择方法、装置、设备及存储介质
CN110555529B (zh) 一种数据处理的方法以及相关装置
CN112742029A (zh) 一种模拟操作的方法、游戏测试的方法以及相关装置
CN113230650B (zh) 一种数据处理方法、装置及计算机可读存储介质
CN112685921B (zh) 一种高效精确搜索的麻将智能决策方法、系统及设备
US20240082728A1 (en) Interaction method and apparatus, and computer storage medium
CN110555480A (zh) 一种训练数据生成的方法及相关装置
CN211025083U (zh) 一种基于单目摄像头手势控制的墙面体感游戏设备
DE112021000598T5 (de) Erweiterbares wörterbuch für spielereignisse
CN111443806A (zh) 交互任务的控制方法、装置、电子设备及存储介质

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: 20868430

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2021563356

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20868430

Country of ref document: EP

Kind code of ref document: A1