CN110152302A - The hands of cards game generate and clustering method, equipment and processor - Google Patents
The hands of cards game generate and clustering method, equipment and processor Download PDFInfo
- Publication number
- CN110152302A CN110152302A CN201810154837.XA CN201810154837A CN110152302A CN 110152302 A CN110152302 A CN 110152302A CN 201810154837 A CN201810154837 A CN 201810154837A CN 110152302 A CN110152302 A CN 110152302A
- Authority
- CN
- China
- Prior art keywords
- hands
- combination
- classification
- processor
- combined
- Prior art date
- Legal status (The legal status 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 status listed.)
- Pending
Links
Classifications
-
- A—HUMAN NECESSITIES
- A63—SPORTS; GAMES; AMUSEMENTS
- A63F—CARD, BOARD, OR ROULETTE GAMES; INDOOR GAMES USING SMALL MOVING PLAYING BODIES; VIDEO GAMES; GAMES NOT OTHERWISE PROVIDED FOR
- A63F13/00—Video games, i.e. games using an electronically generated display having two or more dimensions
- A63F13/60—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
-
- 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
- A63F13/70—Game security or game management aspects
- A63F13/75—Enforcing rules, e.g. detecting foul play or generating lists of cheating players
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
- A63F2300/00—Features of games using an electronically generated display having two or more dimensions, e.g. on a television screen, showing representations related to the game
- A63F2300/50—Features of games using an electronically generated display having two or more dimensions, e.g. on a television screen, showing representations related to the game characterized by details of game servers
- A63F2300/55—Details of game data or player data management
- A63F2300/5586—Details of game data or player data management for enforcing rights or rules, e.g. to prevent foul play
Abstract
The invention discloses a kind of hands of cards game to generate and clustering method, equipment and processor.Wherein, this method comprises: generating hands combination, hands combination includes the hands of cards game each side；Using the model based on machine learning training, hands are combined a classification being determined as in predetermined multiple classifications；And the classification based on hands combination, it is determined whether combined using hands.The present invention solves in cards game the technical issues of appropriate hands combination for how generating " same board power ".
Description
Technical field
The present invention relates to the field of data mining of chess/card game, generate in particular to a kind of hands of cards game
With clustering method, equipment and processor.
Background technique
Two play also referred to as three people's fighting landlords, are a kind of in Chinese 3 popular people's Card Games.On September 3rd, 2016,
Board and Card Games Administrative Center, General Administration of Sport formally starts whole nation sports two and makes a call to a poker tournament (CCPC).In championship
Compound match rule is introduced, i.e., plays identical board and the identical participant in orientation carries out lateral comparison, several secondary bridge queens for all, finally
Ranking is determined by gained total score height.
It is made a call in a match in compound sports two, the fairness for guaranteeing game is combined by using identical hands.However two dozens
The difficult degree of the triumph of one game relies on player and starts the hands obtained before game quality, in order to avoid extreme hands board
Type, existing cards dealing procedure randomly select hands from the specific classification in hands library.This mechanism of dealing out the cards compound two makes a call to one on line
Chance is provided to cheating program in contest.Cheating program can establish mirror image to limited hands library, and generate best board spectrum,
To guarantee that maximum probability is won.
To prevent this kind of " hitting library " cheating.Need to establish one can dynamic random generate " same board power " systems of appropriate hands
System.Therefore for how to establish can dynamic random generate " same board power " systems of appropriate hands, be to guarantee that chess and card movement is fair
Property institute urgent problem.
Aiming at the problem that the appropriate hands combination for how generating " same board power " set forth above, not yet propose at present effective
Solution.
Summary of the invention
Present disclose provides a kind of generation of the hands of cards game and clustering method, equipment and processors, at least to solve
The technical issues of appropriate hands combination of " same board power " certainly how is generated in cards game.
According to one aspect of the disclosure, the hands generation method of cards game is provided, comprising: hands combination is generated,
Hands combination includes the hands of cards game each side；Using the model based on machine learning training, hands combination is determined as pre-
The classification in multiple classifications first determined；And the classification based on hands combination, it is determined whether combined using hands.
The hands clustering method of a kind of cards game another aspect of the present disclosure provides, comprising: be directed to respectively
Multiple hands combined samples extract characteristic value, and hands combination includes the hands of each side of cards game；And based on extracted
The characteristic value of multiple hands combined samples clusters multiple hands combined samples, generates multiple classifications.
A kind of storage medium another aspect of the present disclosure provides, storage medium include the program of storage,
In, program operation when control processor execute any of the above one described in method.
A kind of processor another aspect of the present disclosure provides, processor is for running program, wherein program
Method described in executing any of the above one when operation.
A kind of hands generating device of cards game another aspect of the present disclosure provides, comprising: processor；
And memory, it is connect with processor, for providing the instruction for handling following processing step for processor: hands combination is generated,
Hands combination includes the hands of cards game each side；Using the model based on machine learning training, hands combination is determined as pre-
The classification in multiple classifications first determined；And the classification based on hands combination, it is determined whether combined using hands.
A kind of hands of cards game cluster equipment another aspect of the present disclosure provides, comprising: processor；
And memory, it is connect with processor, for providing the instruction for handling following processing step for processor: being directed to multiple hands respectively
Board combined sample extracts characteristic value, and hands combination includes the hands of each side of cards game；And it is based on extracted multiple hands
The characteristic value of board combined sample clusters multiple hands combined samples, generates multiple classifications.
In embodiment of the disclosure, cluster is carried out by using clustering algorithm crossruff combined sample to generate multiple
Classification, the class for carrying out classification using hands combination of the classifier based on machine learning training to generation and being combined based on hands
Do not judge whether hands combination is the hands combination of same board power, to reach the hands combination for automatically generating same board power
Purpose, and then the technical issues of appropriate hands combination for how generating " same board power ", is solved in cards game.
Detailed description of the invention
The drawings described herein are used to provide a further understanding of the present invention, constitutes part of this application, this hair
Bright illustrative embodiments and their description are used to explain the present invention, and are not constituted improper limitations of the present invention.In the accompanying drawings:
Fig. 1 is for executing showing for the equipment of described in accordance with an embodiment of the present disclosure hands generation method and clustering method
It is intended to；
Fig. 2 is the hands generation method flow chart according to the embodiment of the present disclosure；
Fig. 3 is a kind of schematic diagram of hands combination；
Fig. 4 is the concrete operations flow chart of the step S204 in process shown in Fig. 2；
Fig. 5 is the schematic diagram of the classifier according to used in the embodiment of the present disclosure based on convolutional neural networks；
Fig. 6 is the concrete operations flow chart of the step S404 in process shown in Fig. 4；
Fig. 7 is the flow chart for the method that crossruff combined sample is clustered according to the embodiment of the present disclosure；
Fig. 8 is the schematic diagram of scheme that the combination of the crossruff according to the embodiment of the present disclosure is clustered and classified；
Fig. 9 is the concrete operations flow chart of the step S702 in process shown in Fig. 7；
Figure 10 is a kind of schematic diagram of the discrete probability distribution of the scoring event of hands combination；
Figure 11 is the schematic diagram for dividing score section and extraction characteristic value in different ways；
Figure 12 is the concrete operations flow chart of the step S704 in process shown in Fig. 7；And
Figure 13 A~Figure 13 J is the schematic diagram for 10 classifications that the clustering algorithm according to the embodiment of the present disclosure generates.
Specific embodiment
In order to enable those skilled in the art to better understand the solution of the present invention, below in conjunction in the embodiment of the present invention
Attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is only
The embodiment of a part of the invention, instead of all the embodiments.Based on the embodiments of the present invention, ordinary skill people
The model that the present invention protects all should belong in member's every other embodiment obtained without making creative work
It encloses.
It should be noted that description and claims of this specification and term " first " in above-mentioned attached drawing, "
Two " etc. be to be used to distinguish similar objects, without being used to describe a particular order or precedence order.It should be understood that using in this way
Data be interchangeable under appropriate circumstances, so as to the embodiment of the present invention described herein can in addition to illustrating herein or
Sequence other than those of description is implemented.In addition, term " includes " and " having " and their any deformation, it is intended that cover
Cover it is non-exclusive include, for example, the process, method, system, product or equipment for containing a series of steps or units are not necessarily limited to
Step or unit those of is clearly listed, but may include be not clearly listed or for these process, methods, product
Or other step or units that equipment is intrinsic.
Firstly, the part noun or term that occur during the embodiment of the present disclosure is described are suitable for following solution
It releases:
" hands " described in the disclosure refer to after completing to deal out the cards in cards game, are distributed to the board of each side.Such as
In " fighting landlord " game, after completion is dealt out the cards, the hands of tripartite are respectively 15 boards.
" hands combination " described in the disclosure, refers in cards game, the combination including each side's hands.Such as in " bucket
In landlord " game, hands combination refers to the combination including tripartite's hands.
" same board power " described in the disclosure refers to and is equal in the board power of game each side.
According to embodiments of the present invention, a kind of hands generation method of cards game and the side of hands clustering method are additionally provided
Method embodiment, it should be noted that step shown in the flowchart of the accompanying drawings can be in such as a group of computer-executable instructions
Computer system in execute, although also, logical order is shown in flow charts, in some cases, can be with
Shown or described step is executed different from sequence herein.
Embodiment of the method provided by the embodiment of the present application can be in mobile terminal, terminal or similar operation
It is executed in device.Fig. 1 shows the computer of a kind of the hands generation method for realizing cards game and hands clustering method
The hardware block diagram of terminal (or mobile device).As shown in Figure 1, terminal 10 (or mobile device 10) may include one
(processor 102 may include but unlimited for a or multiple (102a, 102b ... ... being used in figure, 102n is shown) processor 102
In the processing unit of Micro-processor MCV or programmable logic device FPGA etc.), memory 104, Yi Jiyong for storing data
In the transmission module 106 of communication function.In addition to this, it can also include: display, input/output interface (I/O interface), lead to
With the port universal serial bus (USB) (can be used as a port in the port of I/O interface is included), network interface, power supply and/
Or camera.It will appreciated by the skilled person that structure shown in FIG. 1 is only to illustrate, not to above-mentioned electronic device
Structure cause to limit.For example, terminal 10 may also include more perhaps less component or tool than shown in Fig. 1
There is the configuration different from shown in Fig. 1.
It is to be noted that said one or multiple processors 102 and/or other data processing circuits lead to herein
Can often " data processing circuit " be referred to as.The data processing circuit all or part of can be presented as software, hardware, firmware
Or any other combination.In addition, data processing circuit for single independent processing module or all or part of can be integrated to meter
In any one in other elements in calculation machine terminal 10 (or mobile device).As involved in the embodiment of the present application,
The data processing circuit controls (such as the selection for the variable resistance end path connecting with interface) as a kind of processor.
Memory 104 can be used for storing the software program and module of application software, such as the board class in the embodiment of the present invention
Corresponding program instruction/the data storage device of the hands generation method and clustering method of game, processor 102 pass through operation storage
Software program and module in memory 104 are realized above-mentioned thereby executing various function application and data processing
The leak detection method of application program.Memory 104 may include high speed random access memory, may also include nonvolatile memory,
Such as one or more magnetic storage device, flash memory or other non-volatile solid state memories.In some instances, it stores
Device 104 can further comprise the memory remotely located relative to processor 102, these remote memories can be connected by network
It is connected to terminal 10.The example of above-mentioned network includes but is not limited to internet, intranet, local area network, mobile communication
Net and combinations thereof.
Transmitting device 106 is used to that data to be received or sent via a network.Above-mentioned network specific example may include
The wireless network that the communication providers of terminal 10 provide.In an example, transmitting device 106 includes that a network is suitable
Orchestration (Network Interface Controller, NIC), can be connected by base station with other network equipments so as to
Internet is communicated.In an example, transmitting device 106 can be radio frequency (Radio Frequency, RF) module,
For wirelessly being communicated with internet.
Display can such as touch-screen type liquid crystal display (LCD), the liquid crystal display aloow user with
The user interface of terminal 10 (or mobile device) interacts.
Under above-mentioned running environment, this application provides the hands generation methods of cards game as shown in Figure 2.Fig. 2 is
According to the flow chart of the hands generation method of the embodiment of the present disclosure.
Refering to what is shown in Fig. 2, embodiment of the disclosure provides a kind of hands generation method for cards game, need
Bright, step shown in the flowchart of the accompanying drawings can be held in a computer system such as a set of computer executable instructions
Row, although also, logical order is shown in flow charts, and it in some cases, can be to be different from sequence herein
Execute shown or described step.Refering to what is shown in Fig. 2, should include: for the hands generation method of cards game
S202: generating hands combination, and hands combination includes the hands of cards game each side；
S204: using the model based on machine learning training, hands combination is determined as in predetermined multiple classifications
A classification；And
S206: the classification based on hands combination, it is determined whether combined using hands.
To which the hands generation method described in the present embodiment utilizes the model based on machine learning training, to generation
Hands combination is classified, to combine the hands of generation a classification being determined as in multiple classifications.Wherein, multiple classifications
In a part of classification correspond to the hands of " same board power " and combine, another part classification in multiple classifications corresponds to " non-same
The hands combination of equal boards power ".When hands combination is determined as corresponding to " same board power " by the model based on machine learning training
When classification, then combined using this card hand.Otherwise, when hands combination is determined as corresponding to by the model based on machine learning training
When the classification of " non-equally board power ", then combined without using this card hand.
Determine to be combined in this way to the hands of dynamic generation, makes it possible to meet " same board
Power " hands combination used, and when the hands of dynamic generation combine be determined classification it is corresponding be " non-equally board power " when, then
It needs to regenerate hands to combine and repeat above-mentioned method and step.In turn, by utilizing the classification based on machine learning training
Device carries out classification to the hands combination of generation and the classification combined based on hands judges whether hands combination is same board power
Hands combination to achieve the purpose that the hands combination for automatically generating " same board power ", and then solves in cards game such as
What generates the technical issues of appropriate hands combination of " same board power ".
In addition, dynamic generation hands can be combined in a random fashion when generating hands combination.It can also be according to other
Algorithm generates hands combination.And although the technical issues of technical solution of the disclosure is solved is to make a call in a game to find two
The problem of, but after practice, find the method for embodiment of the disclosure for other cards games (including playing card
Cards game in addition) when, it is equally applicable.In addition, model used by the above method can be based on machine learning training
A variety of models, such as can use convolutional neural networks, the other kinds of depth comprising multiple hidden layers can also be used
Practise model.As long as can be by the study and training of sample, by the hands assembled classification of generation to different classifications.
Optionally, model is the classifier based on convolutional neural networks.Convolutional neural networks are that a kind of depth feedforward is artificial
Neural network, according to the practice of inventor, discovery, which carries out classification using the structure of convolutional neural networks come crossruff combination, to be had
Very excellent performance.Specifically, it is illustrated so that cards game two makes a call to one as an example.
Fig. 3 shows two hands combinations for plaing a cards game.Refering to what is shown in Fig. 3, tripartite role in hands combination
Hands be respectively
Role 1 (landlord): 3 " A ", 1 " 2 ", 1 " 3 ", 3 " 4 ", 1 " 5 ", 1 " 6 ", 0 " 7 ", 1 " 8 ", 1
Open " 9 ", 0 " 10 ", 2 " J ", 1 " Q ", 1 " K " and 1 Xiao Wang；
Role 2:0 " A ", 3 " 2 ", 2 " 3 ", 0 " 4 ", 0 " 5 ", 1 " 6 ", 1 " 7 ", 3 " 8 ", 1 " 9 ",
0 " 10 ", 1 " J ", 3 " Q ", 1 " K " and 1 great Wang；
Role 3:1 " A ", 0 " 2 ", 1 " 3 ", 1 " 4 ", 3 " 5 ", 2 " 6 ", 3 " 7 ", 0 " 8 ", 2 " 9 ",
3 " 10 ", 1 " J ", 0 " Q ", 0 " K ", 0 Xiao Wang and 0 great Wang.
Card in one's hand: " K ", " K ", " 10 ".
It is thus possible to indicate hands combination shown in Fig. 3 by 3*15 matrix as shown below:
Further, any hands combination for plaing a game for two can be indicated by 3*15 matrix below:
To can fully describe the hands of 3 players by above-mentioned 3*15 matrix.Then by based on convolution mind
Classifier through network handles above-mentioned 3*15 matrix, the advantage of convolutional neural networks model can further be played
Come.
The output of sorter model based on convolutional neural networks is designed as 1 dimensional vector, and each numerical value indicates defeated in vector
The hands combination entered belongs to the probability of particular category.For example, hands combination is divided into 10 classes in cluster result, then convolutional Neural
Network classifier output is the one-dimensional vector that length is 10.The serial number of model output value is corresponded with classification sequence number.That is, volume
The output result of the sorter model of product neural network is the one-dimensional vector of following form:
(Pc0,Pc1,Pc2,Pc3,Pc4,Pc5,Pc6,Pc7,Pc8,Pc9)。
The board type combination that wherein Pc0 to Pc9 respectively indicates input belongs to the probability of each classification in 10 classifications.
Further, refering to what is shown in Fig. 4, the step S204 of the above method may include following operation:
S402: it is handled by the combination of multiple convolutional layer crossruffs, obtains the first output result；
S404: carrying out sub-sampling processing to the first output result, obtains the second output as a result, wherein the second output result is
One-dimensional vector, and the length of the second output result is equal with the categorical measure of multiple classifications；And
S406: by peripheral sensory neuron layer, the classification of hands combination is determined based on the second output result, wherein first nerves
The quantity for the neuron that first layer includes is equal with the categorical measure of multiple classifications.
Fig. 5 shows the schematic diagram of classifier used in the present embodiment.Refering to what is shown in Fig. 5, used in the present embodiment
Classifier include: multiple convolutional layers 511 to 517, sub-sampling structure 520 and peripheral sensory neuron layer 530.
Multiple convolutional layers are, for example, 7 convolutional layers 511 to 517, and each convolutional layer includes a 3*3 convolution kernel in 128 (or 64)
C1 to C128.Also, after each convolution operation, the matrix size that 0 transmits interlayer is mended around output matrix and is remained unchanged, often
Nonlinear activation function after layer convolution operation selects ELU function.To which the matrix data of hands combination passes through multiple convolutional layers
After 511 to 517 processing, obtain the first output as a result, i.e. convolutional layer output result.
It is a sub- sampling structure 520 after convolutional layer, is exported at result by sub-sampling structure 520 to first
Reason obtains the second output as a result, wherein second exporting result as one-dimensional vector, and the length and multiple classes of the second output result
Other categorical measure is equal.Sub-sampling structure 520 for example can be dropout functional layer structure, is also possible to other sons and adopts
Sample is as a result, preferably stochastical sampling structure.
Also, after sub-sampling structure 520, it is provided with peripheral sensory neuron layer 530.Pass through peripheral sensory neuron layer 530, base
In second output result determine hands combination classification, the quantity for the neuron that wherein peripheral sensory neuron layer 530 includes with it is multiple
The categorical measure of classification is equal.
Wherein, peripheral sensory neuron layer 530 is a full articulamentum, by exporting result for the second of sub-sampling structure 520
Be input to peripheral sensory neuron layer 530, obtain one-dimensional vector as above (Pc0, Pc1, Pc2, Pc3, Pc4, Pc5, Pc6, Pc7, Pc8,
Pc9).The board type combination that wherein Pc0 to Pc9 respectively indicates input belongs to the probability of each classification in 10 classifications.
To which peripheral sensory neuron layer 530 is exported based on second as a result, determining the classification of hands combination.
Optionally, refering to what is shown in Fig. 6, above-mentioned steps S404 further comprises:
S602: sub-sampling processing is carried out to the first output result, chooses the first predetermined quantity from the first output result
Result is exported to be exported as third as a result, wherein the first predetermined quantity is greater than the quantity of classification；
S604: exporting result to third by nervus opticus member layer and handle, and obtains the 4th output as a result, wherein second
Neuronal layers include the neuron of one dimensional arrangement, and the quantity of the neuron of nervus opticus member layer is the first predetermined quantity；With
And
S606: sub-sampling processing is carried out to the 4th output result, selected part output result is made from the 4th output result
For the second output result.
Refering to what is shown in Fig. 5, sub-sampling structure 520 further includes the first sub-sampling structure (i.e. dropout layer 521), the second mind
Through first layer 522 and the second sub-sampling structure (i.e. dropout layer 523).
Dropout layer 521 randomly selects predetermined quantity from the output result (that is, first output result) of convolutional layer structure
(i.e. the first predetermined quantity greater than the quantity of classification, such as can be 64 or output result (that is, third output result) 128).
Nervus opticus member layer 522 is used to handle the export structure (that is, third output result) of dropout layer 521,
The processing result (that is, the 4th output result) of nervus opticus member layer 522 is further obtained, wherein nervus opticus member layer 522 includes
The neuron of one dimensional arrangement, and nervus opticus member layer is full articulamentum, and the quantity of neuron is what dropout layer 521 was chosen
As a result quantity (that is, first predetermined quantity).
Output result (i.e. fourth output result) of the second sub-sampling structure (i.e. dropout layer 523) from neuronal layers 522
Output result (i.e. second output result) of the middle selected part output result as entire sub-sampling structure 520.
As a result, by sub-sampling structure, the scale of input data can be reduced.
Specifically, before classifier used in the disclosure 7 layers be convolutional layer 511 to 517, every layer include 128 (or 64) a 3
The convolution nuclear structure of × 3 sizes after each convolution operation, the matrix size that 0 transmits interlayer is mended around output matrix and is kept
It is constant, and the nonlinear activation function after every layer of convolution operation selects ELU function.
One dropout layer 521 is set after the 7th convolutional layer 517, in model training stage, the dropout layer 521
The part output afferent neuron layer 522 that the 7th convolutional layer 517 can be randomly selected, softens interlayer logic association, enhances the general of model
Change ability.
Neuronal layers 522 are full articulamentum, and the output of the 7th convolutional layer 517 is expanded into one-dimensional vector by dropout layer 521
As the input of neuronal layers 522, neuronal layers 522 include a artificial neuronal structure in 64 (or 256), nonlinear activation function
Still ELU function is selected.
It is a dropout layer 523 after neuronal layers 522, can be selected at random in model training stage dropout layer 523
It takes the part of neuronal layers 522 to export afferent neuron layer 530, softens interlayer logic association, enhance the generalization ability of model.
Neuronal layers 530 are full articulamentum, and neuronal quantity depends on the length of model output one-dimensional vector.For example,
In the case that classification included by the disclosure has 10 classifications, the neuronal layers 530 of 10 class classifiers are full articulamentum, include
10 artificial neuronal structures, the layer do not have nonlinear activation function.
In addition, method described in the present embodiment further includes initializing the parameter of the model by MSRA method.
And method described in the present embodiment further include: pass through following operation training model: the sample of hands combination is obtained
This；Classification processing is carried out to sample using model；Computation model carries out the output result and expected result of classification processing to sample
Between error；And numerical value in the connection weight and convolution kernel of the neuron in model is successively adjusted using Adam algorithm.
Specifically, each parameter is initialized using MSRA method in the convolutional neural networks structure of the disclosure.Model instruction
During white silk, each batch sample data can obtain one group of output by network structure processing, calculate this group output and expectation
After the error of output, entire CNN model is successively adjusted using Adam algorithm (a kind of improved stochastic gradient descent optimization algorithm)
In artificial neuron's connection weight and convolution kernel in numerical value.
To in the above manner, the method for the present embodiment can use hands group of the model to generation of machine learning
Close carry out classification and based on hands combination classification judge hands combination whether be same board power hands combine, to reach
The purpose of the hands combination of same board power is automatically generated, and then solves in cards game and how to generate the suitable of " same board power "
When hands combination the technical issues of.
Further, refering to what is shown in Fig. 7, method described in the present embodiment further includes generating multiple classifications by following operation
S702: characteristic value is extracted for multiple hands combined samples respectively；
S704: the characteristic value based on extracted multiple hands combined samples clusters multiple hands combined samples,
Generate multiple classifications.
Refering to what is shown in Fig. 8, in order to generate the hands of " same board power " combination, according to the sample of hands combination according to institute above
The method stated carries out cluster operation, to obtain the multiple classifications combined about hands.A part of class in plurality of classification
Not Dui Yingyu " same board power " judgement, a part of classification correspond to " non-equally board power " judgement.
Then, it with reference to methods described above, is constructed using multiple classifications that cluster obtains and trains classifier.Example
Such as, as described above, the sample of hands combination is obtained；Classification processing is carried out to sample using model；Computation model to sample into
Error between the output result and expected result of row classification processing；And nerve in model is successively adjusted using Adam algorithm
Numerical value in the connection weight and convolution kernel of member.
Then, using trained classifier, classify to the hands combination of dynamic generation, and according to the hands of generation
Combined classification determines that this card hand combination is the hands combination of " same board power ", the still hands combination of " non-equally board power ".Into
And when determining this card hand combination is the hands combination of " same board power ", it is combined using this card hand and carries out game.Otherwise it gives birth to again
The hands of Cheng Xin combine, and are classified and are determined.
To method through this embodiment, by extracting feature to multiple hands combined samples, and according to the spy of extraction
Sign carries out cluster operation.And then good basis is provided for the training and building of classifier below.Also, since cluster generates
Classification correspond respectively to the judgement of " same board power " or the judgement of " non-equally board power " so that cluster generate multiple classes
It not can be used in the judgement that " same board power " is carried out to the hands combination of dynamic generation.Therefore, by using clustering algorithm opponent
Board combined sample carries out cluster to generate multiple classifications, using the classifier based on machine learning training to the hands group of generation
Close carry out classification and based on hands combination classification judge hands combination whether be same board power hands combine, to reach
The purpose of the hands combination of same board power is automatically generated, and then solves in cards game and how to generate the suitable of " same board power "
When hands combination the technical issues of.
Wherein, refering to what is shown in Fig. 9, the operation for extracting characteristic value includes executing following behaviour for each hands combined sample
Make:
S902: the probability distribution of the scoring event of at least one party in statistics hands combined sample；And
S902: extracting multiple discrete score sections and corresponding probability distribution on the basis of probability distribution, will be multiple
The characteristic value of discrete score section and corresponding probability distribution as hands combined sample.
By taking two play a game as an example, in the case where possessing a large amount of reality to office data, obtains the same board power of hands and divide
The most direct-vision method of standard combines, certain side (such as landlord side) when collecting using this card hand point aiming at various different hands
Several probability distribution, and analyzed, feature is extracted, clustering is carried out.
For example, combining for some hands, the frequency of the discrete score of the landlord occurred, the specific frequency have been counted
Distribution is as shown in Figure 10.Score lower limit is -12288 points, and the score upper limit is 6144 points, totally 95 discrete scores, this distribution is whole
It is approximate for the off-axis normal distribution of symmetry axis with 12 points.
It is combined for this card hand, the discrete score of all possible landlord is divided into several sections, and count each board type
The ratio that reserved portion is distributed in each section.Figure 11 shows the scheme in three kinds of different splitting score sections, respectively
To divide 7 sections, 10 sections and 13 sections.
For being divided into the scheme in 10 sections, then above-mentioned hands combination can be indicated by following characteristics value: (P0,
P1,P2,P3,P4,P5,P6,P7,P8,P9).Wherein P0~P9Respectively this card hand combination under, each section landlord's score it is general
Rate value.To combine, can be indicated by different characteristic values for different hands:
Hands combine 1:(P10,P11,P12,P13,P14,P15,P16,P17,P18,P19)
Hands combine 2:(P20,P21,P22,P23,P24,P25,P26,P27,P28,P29)
……
Hands combine n:(Pn0,Pn1,Pn2,Pn3,Pn4,Pn5,Pn6,Pn7,Pn8,Pn9)。
To complete the extraction of the characteristic value of different hands combinations.And the mapping of characteristic value wherein, is combined to from hands
Unidirectionally to map.That is, correspond only to a characteristic value for a kind of combination of hands, but the same characteristic value but can be with
It is combined corresponding to different hands.
Further, shown in Figure 12, the operation clustered to multiple hands combined samples includes:
S1202: based on the characteristic value extracted from multiple hands combined samples, multiple hands combined samples are carried out tentatively poly-
Class obtains multiple clusters；And
S1204: multiple clusters are further aggregated into multiple classifications.
Wherein preferably, using AP clustering algorithm, by multiple hands combined sample preliminary clusters at multiple clusters.
Specifically, using AP clustering algorithm crossruff combined sample, principle is as described below with carrying out preliminary clusters:
Hands combined sample collection { x is provided1,x2,……,xn}.Using the characteristic value of each sample, every two sample is calculated
Between similarity, and using calculate similarity construct matrix S.Wherein sij>sik, and if only if xiAnd xjSimilitude journey
Degree is greater than itself and xkSimilitude degree.
Then the element of two matrixes R and A are all initialized as zero, and are iterated by following steps:
And
Or
Until decision remains unchanged or more than the number of iterations of setting or a cell after iteration several times
The decision about sample point in domain remains unchanged after iteration for several times, then algorithm terminates.In addition, parameter updates in order to prevent
When vibration, introduce damping parameter lambda.Every information be arranged to its previous iteration updated value λ times adds this information update
The 1- λ times of value.Wherein, attenuation coefficient λ is the real number between 0 to 1.
To by above-described AP algorithm, by multiple hands combined sample preliminary clusters at multiple clusters.
Since the number of cluster after AP cluster is more, the classification of sample is more dispersed, and the sample size for including in cluster is less.For
Next it to use nominal data training board type classifier prepare, need to merge cluster, increase the sample size for including in cluster,
Therefore method provided in this embodiment also merges the result of AP cluster, so that multiple clusters are aggregated into multiple classifications.
It wherein, include: using hierarchical clustering algorithm to multiple clusters by the operation that multiple clusters further aggregate into multiple classifications
It is iterated merging treatment, aggregates into multiple classifications.
Wherein, hierarchical clustering algorithm is chosen the most similar every time on the basis of multiple clusters that AP clustering algorithm generates
Two clusters merge, until merging into a classification.But hierarchical clustering can be shifted to an earlier date by limiting certain index
Cluster is terminated, it is final to achieve the purpose that.
For example, can be using the following conditions as the termination condition of iteration merging treatment: reaching desired classification number in classification
Iteration merging treatment is terminated in the case where amount；Or iteration is terminated in the case where classification reaches desired bulk polymerization degree and is merged
Processing, wherein bulk polymerization degree is measured using the mean value of interior sample extent of polymerization of all categories.
In addition, the operation of preliminary clusters further include: determined according to the different demarcation mode in discrete score section multiple and different
Feature extraction scheme, it is poly- using AP and based on each of multiple and different feature extraction scheme feature extraction scheme
Class algorithm carries out preliminary clusters respectively；And it is multiple poly- to being obtained based on different feature extraction scheme progress preliminary clusters
Class result is assessed respectively, determines optimal feature extraction scheme.And preferably, multiple cluster results are commented respectively
The operation estimated includes that multiple cluster results are utilized respectively with K-L divergence measurement to assess.
Specifically, the similarity degree between the method selection K-L divergence measurement scoring probability distribution sample of the present embodiment,
In, it is distributed by theoretical probability of the probability distribution of sample where cluster centre.Using the flat of samples all in cluster and cluster centre
The extent of polymerization of equal K-L divergence assessment specific clusters；The average value for calculating average K-L divergence in all clusters assesses whole cluster matter
Amount.Average K-L divergence is ckl between remembering cluster, then formula is expressed as follows:
Wherein, si,kIt is the probability distribution of i-th of sample in k-th of cluster, ckIt is the probability distribution at k-th of cluster center, nk
It is the sample size for including in k-th of cluster, m is the number of cluster.
That is, including: using K-L divergence measurement one using the operation that K-L divergence measurement assesses cluster result respectively
All samples are at a distance from cluster center or similarity degree in the cluster of each cluster of a cluster result；Utilize sample in cluster to cluster center
The extent of polymerization of sample in the cluster of one cluster result of average value metric of K-L divergence；And utilize sample in the cluster between each cluster
The mean measures sample cluster bulk polymerization degree of extent of polymerization.
On this basis, the method for the present embodiment is clustered using the sample set of different characteristic extraction scheme, to select
Best features extraction scheme out.Different characteristic extraction scheme corresponds to cluster result and is shown in Table 1:
1 different characteristic extraction scheme of table corresponds to cluster result
Wherein, 7 dimensional feature extraction schemes correspond in cluster result, and AP cluster number of clusters is excessive, and each dimension at part cluster center
It is obviously excessively similar to spend feature, is directly eliminated.
10 dimensional feature extraction schemes correspond in cluster result, and AP clusters number of clusters and 13 dimension schemes are quite and slightly more, and the former
The ckl value of cluster result is smaller than the latter.When the preference parameter using climbing method tuning AP algorithm, (parameter influences
AP clusters the number of clusters of output, no longer sews herein and states), so that AP cluster number of clusters is reached minimum 397,10 dimensional feature extraction sides
The ckl value of case corresponding A P cluster result is still less than the latter.To sum up, the embodiment of the present disclosure has finally chosen is mentioned using 10 dimensional features
It takes scheme and the clustering schemes that AP algorithm preference parameter is -0.8 is set.
Also, certain condition should be followed when adjacent discrete score is merged into score section:
The principle of combined by stages is to enable in each section (hands board type, scoring probability distribution) binary group sample size
Roughly equal, to guarantee going on smoothly for subsequent algorithm process, each cluster sample size is not answered very few；
When selecting in multiple alternative section Merge Scenarios, selection criteria is, after implementing preliminary clusters algorithm, each sample cluster
Quantity and sample cluster bulk polymerization quality, for guarantee hands board type final Clustering Effect it is good enough, sample cluster integrally gathers
It is sufficiently high to close quality needs.
Wherein, the specific choice principle of score section Merge Scenarios includes:
1, select the most important condition of score section Merge Scenarios for the quantity of sample cluster after implementation preliminary clusters algorithm；
2, implement preliminary clusters algorithm after sample cluster quantitative difference it is larger when (extremum be 4 times of average value or more or 1/4 with
Under), the preferential score section Merge Scenarios for selecting sample cluster few；
3, implement preliminary clusters algorithm after sample cluster quantitative difference it is smaller when (extremum be 4 times of average value or less 1/4 with
On), the high score section Merge Scenarios of the preferential bulk polymerization quality for selecting sample cluster.
To determine that the operation of optimal feature extraction scheme includes: to implement in the method provided by the present embodiment
In the case that sample cluster quantitative difference is 4 times or more after preliminary clusters, the few score section Merge Scenarios of selection sample cluster；With
And sample cluster quantitative difference is to select the bulk polymerization degree of sample cluster big in 4 times of situations below after implementing preliminary clusters
Score section Merge Scenarios.
With reference to Figure 13 A-13J, 10 classifications that the clustering according to the present embodiment is formed are explained.By to cluster
As a result 10 classifications of clustering algorithm output can be roughly divided by the image conversion observation of each dimensional characteristics numerical value of middle cluster centre
3 class below.
As the 1st class, Figure 13 A~Figure 13 C shows 3 classifications of landlord's score difficulty.
As the 2nd class, Figure 13 D~Figure 13 G shows easy 4 classifications of landlord's score.
As the 3rd class, Figure 13 H~Figure 13 J shows 3 moderate classifications of landlord's score difficulty.
And wherein, the combination of Figure 13 C and Figure 13 D corresponding hands is very extreme, respectively the board that easily secures satisfactory grades of landlord
Type and landlord easily low point of hands combine, discrimination is poor, is not suitable as very much player's hands.
Therefore, by a classification in above-mentioned 10 classifications, that is, can determine whether the hands assembled classification of dynamic generation
Whether hands combination is the hands combination of " same board power " out, and then determines whether to combine using this card hand and carry out game.
To in embodiment of the disclosure, cluster generating by using clustering algorithm crossruff combined sample
Multiple classifications classify and combine based on hands using hands combination of the classifier based on machine learning training to generation
Classification judge hands combination whether be same board power hands combination, to reach the hands group for automatically generating same board power
The purpose of conjunction, and then the technical issues of appropriate hands combination for how generating " same board power ", is solved in cards game.
And further, the cluster output of statistical method is being compared with other algorithms used in the present embodiment
When need the form of unified cluster output: taking all kinds of board types is cluster to the mean value interval intermediate value and variance section intermediate value of reserved portion
Theoretical center point.Then the Clustering Theory center is corrected, searches for expectation and standard deviation of each board type to reserved portion, takes distance
The nearest board type of statistical method Clustering Theory central point-feature samples are used as practical cluster centre.
Under the premise of determining number of clusters, to statistical method, classical K-Means clustering algorithm and cluster calculation stage by stage
The clustering result quality of method is assessed respectively, and the results are shown in Table 1.
The assessment of 1 sorting algorithm of table
It can be seen that clustering algorithm is poly- in the initial board type of fighting landlord according to hands quality stage by stage provided by the present embodiment
There is significant advantage in class scene.
Further, the present embodiment additionally provides a kind of hands clustering method for cards game.Refering to what is shown in Fig. 5,
This method comprises:
S502: extracting characteristic value for multiple hands combined samples respectively, and hands combination includes each side of cards game
Hands；And
S504: the characteristic value based on extracted multiple hands combined samples clusters multiple hands combined samples,
Generate multiple classifications.
Optionally, the operation for extracting characteristic value includes that following operation is executed for each hands combined sample: statistics hand
The probability distribution of the scoring event of at least one party in board combined sample；Multiple discrete scoring areas are extracted on the basis of probability distribution
Between and corresponding probability distribution, using multiple discrete score sections and corresponding probability distribution as the spy of hands combined sample
Value indicative.
Optionally, the operation clustered to multiple hands combined samples includes: to be based on mentioning from multiple hands combined samples
The characteristic value taken carries out preliminary clusters to multiple hands combined samples, obtains multiple clusters；And multiple clusters are further aggregated into
Multiple classifications.
Optionally, carry out preliminary clusters operation include: based on extracted characteristic value, will be more using AP clustering algorithm
A hands combined sample preliminary clusters are at multiple clusters.
Optionally, multiple clusters are further aggregated into the operation of multiple classifications includes: using hierarchical clustering algorithm to multiple
Cluster is iterated merging treatment, aggregates into multiple classifications.
Optionally, using the following conditions as the termination condition of iteration merging treatment: reaching desired categorical measure in classification
In the case where terminate iteration merging treatment；Or it is terminated at iteration merging in the case where classification reaches desired bulk polymerization degree
Reason, wherein bulk polymerization degree is measured using the mean value of interior sample extent of polymerization of all categories.
Optionally, the operation of preliminary clusters further include: according to the different demarcation mode in discrete score section determine it is multiple not
Same feature extraction scheme, and based on each of multiple and different feature extraction scheme feature extraction scheme, utilize AP
Clustering algorithm carries out preliminary clusters respectively；And it is multiple to being obtained based on different feature extraction scheme progress preliminary clusters
Cluster result is assessed respectively, determines optimal feature extraction scheme.
Optionally, the operation assessed respectively multiple cluster results includes being utilized respectively K-L to multiple cluster results
Divergence measurement is assessed.
It optionally, include: using K-L divergence degree using the operation that K-L divergence measurement assesses cluster result respectively
All samples are measured in the cluster of each cluster an of cluster result at a distance from cluster center or similarity degree；Utilize sample in cluster to cluster
The extent of polymerization of sample in the cluster of one cluster result of average value metric of center K-L divergence；And using in the cluster between each cluster
The mean measures sample cluster bulk polymerization degree of sample extent of polymerization.
Optionally it is determined that the operation of optimal feature extraction scheme include: after implementing preliminary clusters sample number of clusters amount it is poor
Wei not be in the case where 4 times or more, the score section Merge Scenarios that select sample cluster few；And the sample after implementing preliminary clusters
Cluster quantitative difference is the big score section Merge Scenarios of the bulk polymerization degree of selection sample cluster in 4 times of situations below.
It should be noted that for the various method embodiments described above, for simple description, therefore, it is stated as a series of
Combination of actions, but those skilled in the art should understand that, the present invention is not limited by the sequence of acts described because
According to the present invention, some steps may be performed in other sequences or simultaneously.Secondly, those skilled in the art should also know
It knows, the embodiments described in the specification are all preferred embodiments, and related actions and modules is not necessarily of the invention
It is necessary.
Through the above description of the embodiments, those skilled in the art can be understood that according to above-mentioned implementation
The method of example can be realized by means of software and necessary general hardware platform, naturally it is also possible to by hardware, but it is very much
In the case of the former be more preferably embodiment.Based on this understanding, technical solution of the present invention is substantially in other words to existing
The part that technology contributes can be embodied in the form of software products, which is stored in a storage
In medium (such as ROM/RAM, magnetic disk, CD), including some instructions are used so that a terminal device (can be mobile phone, calculate
Machine, server or network equipment etc.) method that executes each embodiment of the present invention.
In the above embodiment of the invention, it all emphasizes particularly on different fields to the description of each embodiment, does not have in some embodiment
The part of detailed description, reference can be made to the related descriptions of other embodiments.
In several embodiments provided herein, it should be understood that disclosed technology contents can pass through others
Mode is realized.Wherein, the apparatus embodiments described above are merely exemplary, such as the division of the unit, only
A kind of logical function partition, there may be another division manner in actual implementation, for example, multiple units or components can combine or
Person is desirably integrated into another system, or some features can be ignored or not executed.Another point, shown or discussed is mutual
Between coupling, direct-coupling or communication connection can be through some interfaces, the INDIRECT COUPLING or communication link of unit or module
It connects, can be electrical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme
's.
It, can also be in addition, the functional units in various embodiments of the present invention may be integrated into one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.Above-mentioned integrated list
Member both can take the form of hardware realization, can also realize in the form of software functional units.
If the integrated unit is realized in the form of SFU software functional unit and sells or use as independent product
When, it can store in a computer readable storage medium.Based on this understanding, technical solution of the present invention is substantially
The all or part of the part that contributes to existing technology or the technical solution can be in the form of software products in other words
It embodies, which is stored in a storage medium, including some instructions are used so that a computer
Equipment (can for personal computer, server or network equipment etc.) execute each embodiment the method for the present invention whole or
Part steps.And storage medium above-mentioned includes: that USB flash disk, read-only memory (ROM, Read-Only Memory), arbitrary access are deposited
Reservoir (RAM, Random Access Memory), mobile hard disk, magnetic or disk etc. be various to can store program code
Medium.
The above is only a preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For member, various improvements and modifications may be made without departing from the principle of the present invention, these improvements and modifications are also answered
It is considered as protection scope of the present invention.
Claims (10)
1. a kind of hands generation method of cards game characterized by comprising
Hands combination is generated, the hands combination includes the hands of the cards game each side；
Using the model based on machine learning training, one that hands combination is determined as in predetermined multiple classifications
Classification；And
Classification based on hands combination, it is determined whether combined using the hands.
2. the method according to claim 1, wherein the model is the classifier based on convolutional neural networks.
3. according to the method described in claim 2, it is characterized in that, hands combination is determined as using the classifier
The operation of a classification in predetermined multiple classifications, comprising:
Hands combination is handled by multiple convolutional layers, obtains the first output result；
Sub-sampling processing is carried out to the first output result, obtains the second output as a result, wherein the second output result is
One-dimensional vector, and the length of the second output result is equal with the categorical measure of the multiple classification；And
By peripheral sensory neuron layer, the classification of the hands combination is determined based on the second output result, wherein described first
The quantity for the neuron that neuronal layers include is equal with the categorical measure of the multiple classification.
4. the method according to claim 1, wherein the method also includes by following operation, described in generation
Multiple classifications:
Characteristic value is extracted for multiple hands combined samples respectively；And
Based on the characteristic value of extracted the multiple hands combined sample, the multiple hands combined sample is clustered,
Generate the multiple classification.
5. according to the method described in claim 4, it is characterized in that, the operation for extracting characteristic value includes being directed to each hand
Board combined sample executes following operation:
Count the probability distribution of the scoring event of at least one party in hands combined sample；
Multiple discrete score sections and corresponding probability distribution are extracted on the basis of the probability distribution, by it is the multiple from
Dissipate the characteristic value of score section and corresponding probability distribution as hands combined sample.
6. the hands clustering method of cards game characterized by comprising
Characteristic value is extracted for multiple hands combined samples respectively, the hands combination includes the hand of each side of the cards game
Board；And
Based on the characteristic value of extracted the multiple hands combined sample, the multiple hands combined sample is clustered,
Generate multiple classifications.
7. a kind of storage medium, which is characterized in that the storage medium includes the program of storage, wherein
The method described in any one of control processor perform claim requirement 1 to 6 in described program operation.
8. a kind of processor, which is characterized in that the processor is for running program, wherein right of execution when described program is run
Benefit require any one of 1 to 6 described in method.
9. a kind of hands generating device of cards game characterized by comprising
Processor；And
Memory is connected to the processor, for providing the instruction for handling following processing step for the processor:
Hands combination is generated, the hands combination includes the hands of the cards game each side；
Using the model based on machine learning training, one that hands combination is determined as in predetermined multiple classifications
Classification；And
Classification based on hands combination, it is determined whether combined using the hands.
10. the hands of cards game cluster equipment characterized by comprising
Processor；And
Memory is connected to the processor, for providing the instruction for handling following processing step for the processor:
Characteristic value is extracted for multiple hands combined samples respectively, the hands combination includes the hand of each side of the cards game
Board；And
Based on the characteristic value of extracted the multiple hands combined sample, the multiple hands combined sample is clustered,
Generate multiple classifications.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810154837.XA CN110152302A (en) | 2018-02-23 | 2018-02-23 | The hands of cards game generate and clustering method, equipment and processor |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810154837.XA CN110152302A (en) | 2018-02-23 | 2018-02-23 | The hands of cards game generate and clustering method, equipment and processor |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110152302A true CN110152302A (en) | 2019-08-23 |
Family
ID=67635602
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810154837.XA Pending CN110152302A (en) | 2018-02-23 | 2018-02-23 | The hands of cards game generate and clustering method, equipment and processor |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110152302A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110548277A (en) * | 2019-09-29 | 2019-12-10 | 腾讯科技（深圳）有限公司 | Method, device and equipment for acquiring hand cards in card game program and readable medium |
| CN110598853A (en) * | 2019-09-11 | 2019-12-20 | 腾讯科技（深圳）有限公司 | Model training method, information processing method and related device |
| CN111185010A (en) * | 2019-12-25 | 2020-05-22 | 北京理工大学 | System and method for constructing landlord card-playing program by using pulse neural network |
| CN113413587A (en) * | 2021-07-02 | 2021-09-21 | 腾讯科技（深圳）有限公司 | Information determination method, device, equipment and medium for card sports |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090018981A1 (en) * | 2007-07-13 | 2009-01-15 | Microsoft Corporation | Learning classifiers using combined boosting and weight trimming |
| TW201134529A (en) * | 2010-04-12 | 2011-10-16 | Astro Corp | Method of poker game, computer program product, and game device |
| CN102258865A (en) * | 2010-05-27 | 2011-11-30 | 北京联众电脑技术有限责任公司 | Composite team Chinese poker card game on internet |
| CN105771237A (en) * | 2014-12-16 | 2016-07-20 | 博雅网络游戏开发（深圳）有限公司 | Method and device for card changing control in chess and card games |
| CN106295541A (en) * | 2016-08-03 | 2017-01-04 | 乐视控股（北京）有限公司 | Vehicle type recognition method and system |
| CN107330452A (en) * | 2017-06-16 | 2017-11-07 | 悦享趋势科技（北京）有限责任公司 | Clustering method and device |
| CN107633007A (en) * | 2017-08-09 | 2018-01-26 | 五邑大学 | A kind of comment on commodity data label system and method based on stratification AP clusters |
-
2018
- 2018-02-23 CN CN201810154837.XA patent/CN110152302A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090018981A1 (en) * | 2007-07-13 | 2009-01-15 | Microsoft Corporation | Learning classifiers using combined boosting and weight trimming |
| TW201134529A (en) * | 2010-04-12 | 2011-10-16 | Astro Corp | Method of poker game, computer program product, and game device |
| CN102258865A (en) * | 2010-05-27 | 2011-11-30 | 北京联众电脑技术有限责任公司 | Composite team Chinese poker card game on internet |
| CN105771237A (en) * | 2014-12-16 | 2016-07-20 | 博雅网络游戏开发（深圳）有限公司 | Method and device for card changing control in chess and card games |
| CN106295541A (en) * | 2016-08-03 | 2017-01-04 | 乐视控股（北京）有限公司 | Vehicle type recognition method and system |
| CN107330452A (en) * | 2017-06-16 | 2017-11-07 | 悦享趋势科技（北京）有限责任公司 | Clustering method and device |
| CN107633007A (en) * | 2017-08-09 | 2018-01-26 | 五邑大学 | A kind of comment on commodity data label system and method based on stratification AP clusters |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110598853A (en) * | 2019-09-11 | 2019-12-20 | 腾讯科技（深圳）有限公司 | Model training method, information processing method and related device |
| CN110598853B (en) * | 2019-09-11 | 2022-03-15 | 腾讯科技（深圳）有限公司 | Model training method, information processing method and related device |
| CN110548277A (en) * | 2019-09-29 | 2019-12-10 | 腾讯科技（深圳）有限公司 | Method, device and equipment for acquiring hand cards in card game program and readable medium |
| CN110548277B (en) * | 2019-09-29 | 2020-12-25 | 腾讯科技（深圳）有限公司 | Method, device and equipment for acquiring hand cards in card game program and readable medium |
| CN111185010A (en) * | 2019-12-25 | 2020-05-22 | 北京理工大学 | System and method for constructing landlord card-playing program by using pulse neural network |
| CN113413587A (en) * | 2021-07-02 | 2021-09-21 | 腾讯科技（深圳）有限公司 | Information determination method, device, equipment and medium for card sports |
| CN113413587B (en) * | 2021-07-02 | 2022-11-15 | 腾讯科技（深圳）有限公司 | Information determination method, device, equipment and medium for card sports |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110152302A (en) | The hands of cards game generate and clustering method, equipment and processor | |
| CN109508789A (en) | Predict method, storage medium, processor and the equipment of hands | |
| CN109814955B (en) | Method and device for determining formation capacity and electronic equipment | |
| CN109513215A (en) | A kind of object matching method, model training method and server | |
| CN107970608A (en) | The method to set up and device, storage medium, electronic device of outpost of the tax office game | |
| CN112016704B (en) | AI model training method, model using method, computer device and storage medium | |
| CN109603159A (en) | Match the method, apparatus and system of game player | |
| CN108319974A (en) | Data processing method, device, storage medium and electronic device | |
| CN109999496A (en) | Control method, device and the electronic device of virtual objects | |
| CN110152304A (en) | Determination method and device, storage medium and the electronic device of triumph value | |
| CN108671546A (en) | Determination method and apparatus, storage medium and the electronic device of object run | |
| CN108304853A (en) | Acquisition methods, device, storage medium and the electronic device for the degree of correlation of playing | |
| CN107341548A (en) | A kind of data processing method, device and electronic equipment | |
| CN111957047B (en) | Checkpoint configuration data adjustment method, computer equipment and storage medium | |
| CN110232405A (en) | Method and device for personal credit file | |
| CN110339563A (en) | The generation method and device of virtual objects, storage medium and electronic device | |
| CN110008999A (en) | Determination method, apparatus, storage medium and the electronic device of target account number | |
| CN109453524A (en) | A kind of method of object matching, the method for model training and server | |
| CN109718558A (en) | The determination method and apparatus of game information, storage medium, electronic device | |
| CN109011580A (en) | Final phase of a chess game board face acquisition methods, device, computer equipment and storage medium | |
| Ashlock et al. | General video game playing escapes the no free lunch theorem | |
| Lockett et al. | Evolving explicit opponent models in game playing | |
| CN111701240B (en) | Virtual article prompting method and device, storage medium and electronic device | |
| CN108771869A (en) | Performance test methods and device, storage medium and electronic device | |
| CN108764985A (en) | Beat reward method, apparatus and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| AD01 | Patent right deemed abandoned |
Effective date of abandoning: 20230516 |
|
| AD01 | Patent right deemed abandoned |