Electronic game providing system, electronic game providing method, and electronic game providing program

Abstract

To provide an artificial intelligence agent capable of dynamically adjusting strength in an electronic game.SOLUTION: A game server 106 includes an artificial intelligence agent 62 that has learned so as to use at least game information 64 including information related to game progress in an electronic game and a strength index indicating strength in the electronic game and output a strategy of playing the electronic game with strength in accordance with the strength index.SELECTED DRAWING: Figure 6

Description

[Technical Field]

[0001] The present invention relates to an electronic game providing system, an electronic game providing method, and an electronic game providing program. [Background technology]

[0002] Conventionally, information processing devices that provide electronic games are known. Known electronic games provided by such information processing devices include electronic games that use multiple objects (in-game objects) such as characters and cards. One example is a competitive game in which a player uses multiple objects to battle against an opponent, either a computer or another player.

[0003] Meanwhile, in recent years, research into machine learning has been actively conducted. For example, research is being conducted into deep learning using multi-layer neural networks. Using such deep learning, artificial intelligence (AI) agents that play electronic games are trained, and AI agents with higher performance (i.e., better at games) than conventional ones have been realized. For example, a technology for efficiently training an AI agent that plays an electronic game using multiple objects has been disclosed (Patent Document 1). [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent Publication No. 2019-95973 Summary of the Invention [Problem to be solved by the invention]

[0005] However, in conventional AI agents using AI, the strength of the agents had to be adjusted manually by the planner or the player themselves, which meant that the strength of the AI ​​agents in electronic games could not be quantitatively optimized.

[0006] One of the objects of the present invention is to provide an artificial intelligence agent whose strength in an electronic game can be dynamically adjusted. [Means for solving the problem]

[0007] One aspect of the present invention is an electronic game provision system characterized by including an artificial intelligence agent that is trained to use at least game information including information related to game progress in an electronic game and a strength index indicating strength in the electronic game, and to output a strategy for playing the electronic game with a strength corresponding to the strength index.

[0008] Here, it is preferable that the artificial intelligence agent is trained by using a reward database that associates the strength index with the winning rate in the electronic game, and by giving rewards to each artificial intelligence agent in a match between the artificial intelligence agents so that the winning rate corresponds to the strength index input to each artificial intelligence agent.

[0009] In addition, in a battle between the AI ​​agents, it is preferable that the strength index input to each AI agent be set randomly.

[0010] It is also preferable that the electronic game is provided with an estimator that estimates a player strength index that indicates the strength of the player when playing the electronic game, and that when playing the electronic game using the artificial intelligence agent, the strength index input to the artificial intelligence agent is set in accordance with the player strength index.

[0011] It is also preferable that the estimator estimates the player strength index based on a progress history of the player's play of the electronic game.

[0012] The estimator preferably estimates the player strength index based on the results of the player's play of the electronic game. For example, the estimator preferably estimates the player strength index based on the statistical winning rate of the player's play of the electronic game.

[0013] It is also preferable that the estimator includes an artificial intelligence estimator that has been trained to receive input of historical information from when a player played the electronic game and output the player's strength in the electronic game.

[0014] Another aspect of the present invention is an electronic game providing program that causes a computer to function as an artificial intelligence agent that has been trained to use at least game information including information related to the progress of an electronic game and a strength index indicating strength in the electronic game, and to output a strategy for playing the electronic game with a strength corresponding to the strength index.

[0015] Another aspect of the present invention is a method for providing an electronic game by causing a computer to function as an artificial intelligence agent that is trained to use at least game information including information related to game progress in the electronic game and a strength index indicating strength in the electronic game, and to output a strategy for playing the electronic game with a strength corresponding to the strength index.

[0016] Another aspect of the present invention is a method for providing an electronic game, characterized in that a computer is trained as an artificial intelligence agent that uses at least game information including information related to the game progress in an electronic game and a strength index indicating strength in the electronic game, and outputs a strategy for playing the electronic game with strength corresponding to the strength index. [Effects of the Invention]

[0017] According to the present invention, it is possible to provide an artificial intelligence agent whose strength in an electronic game can be dynamically adjusted. Other objects of the embodiments of the present invention will become apparent by reading the entire specification. [Brief explanation of the drawings]

[0018] [Figure 1] 1 is a diagram showing a configuration of an electronic game providing system according to an embodiment of the present invention. [Figure 2] FIG. 2 is a diagram illustrating a configuration of a planner terminal according to an embodiment of the present invention. [Figure 3] FIG. 2 is a diagram showing a configuration of a player terminal according to an embodiment of the present invention. [Figure 4] FIG. 2 is a diagram showing the configuration of a game server according to an embodiment of the present invention. [Figure 5] FIG. 10 is a diagram showing an example of a game screen of an electronic game. [Figure 6] FIG. 2 is a functional block diagram showing functions of a game server according to an embodiment of the present invention. [Figure 7] FIG. 2 is a diagram illustrating a learning method for an artificial intelligence agent according to an embodiment of the present invention. [Figure 8] FIG. 2 is a diagram illustrating an example of a reward database according to an embodiment of the present invention. [Figure 9] 1 is a diagram showing a method for playing an electronic game in an embodiment of the present invention. FIG. [Figure 10] 10A and 10B are diagrams illustrating a learning method for a strength index estimator according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0019] 1, an electronic game providing system 100 according to an embodiment of the present invention includes a planner terminal 102, a player terminal 104, and a game server 106. The planner terminal 102 and the game server 106, and the player terminal 104 and the game server 106 are connected to each other so as to be able to communicate with each other via an information and communication network 108 such as the Internet or a LAN.

[0020] In the electronic game providing system 100, a planner can develop an artificial intelligence agent for playing an electronic game by accessing a game server 106 using a planner terminal 102. Then, in response to a request for providing an electronic game transmitted from a player terminal 104, the game server 106 transmits game information relating to the progress of the game to the player terminal 104, thereby providing the electronic game.

[0021] The planner terminal 102 is a terminal used by a planner, who is a developer of an electronic game provided by the electronic game providing system 100. The planner terminal 102 can be configured, for example, as a general-purpose computer. As shown in FIG. 2, the planner terminal 102 includes a processing unit 10, a memory unit 12, an input unit 14, an output unit 16, and a communication unit 18. The processing unit 10 includes a means for performing calculation processing, such as a CPU. The processing unit 10 realizes the function of a planner terminal in the electronic game providing system 100 of this embodiment by executing a planner terminal program (e.g., a program for functioning as an operation system) stored in the memory unit 12. The memory unit 12 includes storage means, such as a semiconductor memory or a memory card. The memory unit 12 is accessible and connected to the processing unit 10, and stores the planner terminal program and information necessary for its processing. The input unit 14 includes a means for inputting information. The input unit 14 includes, for example, a keyboard, a touch panel, buttons, etc., for receiving input from the planner. The output unit 16 includes means for outputting the processing results of the planner terminal 102, such as a user interface screen (UI) for receiving input information from the planner. The output unit 16 is equipped with, for example, a display for presenting images to the planner. The communication unit 18 is configured to include an interface for communicating information with the game server 106 via the information communication network 108. Communication by the communication unit 18 may be wired or wireless.

[0022] The player terminal 104 is a terminal used by a player to play an electronic game provided by the game server 106. The player terminal 104 may be, for example, a mobile terminal such as a smartphone or a tablet terminal, but is not limited thereto and may also be a stationary computer or the like. As shown in FIG. 3 , the player terminal 104 includes a processing unit 20, a memory unit 22, an input unit 24, an output unit 26, and a communication unit 28. The processing unit 20 includes a means for performing arithmetic processing, such as a CPU. The processing unit 20 realizes the function of providing an electronic game to a player by executing a client program (or client application) stored in the memory unit 22. Specifically, the player terminal 104 functions as a so-called electronic game terminal. The memory unit 22 includes storage means such as a semiconductor memory or a memory card. The memory unit 22 is accessible and connected to the processing unit 20 and stores the client program and information required for its processing. The input unit 24 includes means for inputting information. The input unit 24 includes, for example, a keyboard, a touch panel, console buttons, etc., for receiving input from the player. The output unit 26 includes a means for displaying a game screen when providing an electronic game, a user interface screen (UI) for receiving input information from a player, and the like. The output unit 26 is equipped with, for example, a display that presents images to a player. The communication unit 28 is configured to include an interface that communicates information with the game server 106 via the information communication network 108. Communication by the communication unit 28 may be wired or wireless.

[0023] The game server 106 is a server that provides electronic games to the player terminals 104. The game server 106 can be realized, for example, by a general-purpose computer. As shown in FIG. 4, the game server 106 includes a processing unit 30, a memory unit 32, an input unit 34, an output unit 36, and a communication unit 38. The processing unit 30 includes a means for performing arithmetic processing, such as a CPU. The processing unit 30 executes an electronic game providing program stored in the memory unit 32 to provide the electronic game in the electronic game providing system 100 of this embodiment. In this embodiment, the game server 106 provides the player terminals 104 with information necessary for progressing through the electronic game and also functions as an artificial intelligence agent that controls the actions of non-player characters (NPCs) in the electronic game. The memory unit 32 includes storage means such as a semiconductor memory or a memory card. The memory unit 32 is accessible to the processing unit 30 and stores the electronic game providing program, information for functioning as an artificial intelligence agent, and other information necessary for processing the electronic game. The input unit 34 includes a means for inputting information. The input unit 34 includes, for example, a keyboard, a touch panel, buttons, etc. that receive input from the user. The output unit 36 ​​includes means for outputting information necessary for processing in the game server 106, such as a user interface screen (UI) for receiving input information from the user. The output unit 36 ​​includes, for example, a display that presents images to the user. The communication unit 38 includes an interface that communicates information with the planner terminal 102 and the player terminal 104 via the information communication network 108. Communication by the communication unit 38 may be wired or wireless.

[0024] In this embodiment, the planner terminal 102, the player terminal 104, and the game server 106 are implemented as separate terminals, but some or all of the processing of these terminals may be implemented on a single terminal, or may be distributed and implemented on further terminals.

[0025] An example of an electronic game used as an example in this embodiment will now be described. Fig. 5 shows an example of a game screen displayed on the player terminal 104. In the electronic game, a player uses a plurality of characters as a plurality of objects to compete against an opponent player. The opponent player may be a player other than the player himself, or may be a computer (game server 106). Specifically, the player himself and the opponent player alternately place their characters on a board 200 on which a plurality of squares are defined as a game field, in the manner of Othello (registered trademark). In Fig. 5, the rows of each square are indicated by numbers "1" to "6" and the columns by letters "A" to "F."

[0026] White circles on the board 200 indicate the player's stones (white stones), and black circles indicate the opponent player's stones (black stones). White circles on the board 200 with character names (e.g., "C0055") written inside them indicate the player's characters (player characters), and black circles with character names written inside them indicate the opponent player's characters (enemy characters). When an enemy character or black stone is sandwiched between two player characters, or between one white stone and one player character, the enemy character or black stone can be transformed (flipped) into a white stone. When a player's character or white stone is sandwiched between two enemy characters, or between one black stone and one enemy character, the player's character or white stone is transformed into a black stone. As in regular Othello, at the start of the game, two white stones and two black stones may be placed in the four central squares of the board 200. That is, white stones may be placed on the squares 3C and 4D, and black stones may be placed on the squares 3D and 4C.

[0027] A player selects multiple characters from among the multiple characters defined in the electronic game and acquired by the player (available in the electronic game). For example, in this embodiment, 16 characters are selected. The group of 16 characters selected by the player is called a deck. The player can acquire new characters as the game progresses or by paying a fee. As shown in FIG. 5, four characters selected from the deck are displayed on the screen. The four characters are called hand pieces 202. The hand pieces 202 may be randomly selected from the deck by the game server 106, or the player may be able to select them from the deck. The user selects any character from the four hand pieces 202 and places the selected character on a square on the board 200. When one character from the hand pieces 202 is placed on the board 200, one character selected from the deck is added to the hand pieces 202.

[0028] Each character is set with multiple parameters, such as attack power, skill effect, and skill activation conditions. When a character is placed on the board 200, it can inflict damage on an opponent according to the character's attack power. Alternatively, when the character's skill activation conditions are met, a skill (technique) as a special effect possessed by the character is executed, and a game effect according to the skill effect (or skill effect and attack power) is activated.

[0029] Each player and opponent player has a set HP (hit points). In FIG. 5, the player's maximum HP is 250 and the opponent's current HP is 180. The opponent's HP can be reduced by placing a character on the board 200 or by using a skill. The character's skill may also be able to restore HP. The player whose HP reaches 0 first loses, and the other player wins.

[0030] However, the electronic games provided by the electronic game providing system 100 are not limited to the above electronic games, but may be any electronic games in which a player competes against a non-player character (NPC) controlled by an artificial intelligence agent provided by the game server 106.

[0031] Fig. 6 is a functional block diagram showing the processing executed in the game server 106. The game server 106 realizes an artificial intelligence agent using the functions shown in Fig. 6, and when providing an electronic game, the artificial intelligence agent automatically controls a non-player character (NPC), allowing the player to play against the NPC.

[0032] As described above, characters are used in electronic games. Information about characters used in electronic games is stored as a character DB 60 in the storage unit 32. FIG. 7 shows an example of the character DB 60. In FIG. 7, one record corresponds to one character. In this embodiment, each character has a character name, attributes that indicate the characteristics of the character, a rarity that indicates the difficulty of obtaining the character, an attack power that is the basis for the damage that the character will inflict on an opponent when placed on the board 200, a skill name, a skill activation condition that indicates the conditions under which the character's skill is activated, and skill effect parameters that are the effect when the skill is activated. Note that "normal attack" in the skill effect column means that the character is placed on the board 200 and inflicts damage on an opponent based on the attack power value without activating a skill.

[0033] The control unit 30 functions to train an artificial intelligence agent through machine learning using various information stored in the memory unit 32, and to automatically operate non-player characters (NPCs) using the trained artificial intelligence agent.

[0034] As shown in FIG. 6, the control unit 30 functions as a learning unit 50, an AI battle processing unit 52, a strength index estimator 54, and a player log collection unit 56.

[0035] The learning unit 50 trains the AI ​​agent using game information 64, which includes information about the features of characters used in the electronic game and the situation of the electronic game. The memory unit 32 stores the AI ​​agent 62 obtained by training. The AI ​​agent 62 is trained using a method that uses deep learning. The AI ​​agent 62 is configured to include, for example, a multilayer neural network. The multilayer neural network is defined by various parameters (such as the layer structure, the neuron structure in each layer, and the weight or bias of each neuron). Therefore, the memory unit 32 stores various parameters related to the network that constitutes the model for the AI ​​agent 62 (such as the layer structure, the neuron structure in each layer, and the weight coefficient or bias of each neuron).

[0036] The learning unit 50 trains the artificial intelligence agent so that, when it receives the feature quantities of the characters used in the electronic game and game information 64 related to the electronic game, it outputs a strategy for using the characters in the electronic game. The strategy for using the characters is a strategy that includes what features should be used in the electronic game, how the characters should be used, etc.

[0037] For example, in the example of the electronic game described above, the strategy includes operations that the player needs to perform to progress the game, such as how to arrange the pieces 202 on the deck, how to select the pieces 202 to use from the pieces 202 arranged on the deck, and where to arrange the selected pieces 202 on the board 200.

[0038] The game information 64 includes feature quantities of characters used in the electronic game. The game information 64 also includes information indicating the game situation at each point in time during the progress of the electronic game. Note that the game situation at each point in time during the progress of the electronic game may include not only the current game situation but also past game situations (history) leading up to the current situation.

[0039] For example, in the case of an electronic game in which players alternately place characters on the board 200, the game information 64 may include information such as the player's and opponent's stones placed on the board 200, their characters, the hand pieces 202 placed in the deck, the stones (characters) held by the players, the player's maximum HP, and current HP. Specifically, the game information 64 may include the following items: "number of turns," "placed characters" indicating the characters placed on the board 200 in that turn, "placement location" indicating the location where the characters were placed in that turn, "number of stones flipped over" indicating the number of opponent's stones or enemy characters flipped over in that turn, "skill activation" indicating whether a skill of a character placed in that turn was activated, and "skill effect value" indicating the effect (e.g., amount of damage) caused by the skill activation. Furthermore, the game information 64 may include other items related to the progress of the electronic game.

[0040] The character feature may be an expression vector for the character. The character expression vector can be obtained using an expression vector extractor that has learned feature vectors representing the features of each character registered in the character DB 60. The expression vector extractor can be realized by learning a distributed representation of each character used in the electronic game. For example, the expression vector extractor can be obtained by using a representation learning learner consisting of a neural network to learn the values ​​of each element of the character's feature vector based on the game state before the character was used (such as the placement of stones or characters on the board 200) and the result of using the character (the next state). In the above example of the electronic game, the character's feature vector is, for example, a five-dimensional vector, with each element representing a strategic feature in the electronic game (e.g., "desired to place it in a corner" or "desired to use it early in the game"). In this way, by using a distributed representation of each character, it is possible to represent multiple characters using lower-dimensional vectors. Furthermore, characters with similar strategic features in the electronic game will be represented by similar feature vectors. Various known learning methods can be used for representation learning.

[0041] By representing characters as feature vectors and then training the AI ​​agent 62, the amount of calculation required for training can be reduced compared to when each character is individually labeled. Furthermore, because characters with similar features are represented by similar feature vectors, training can be carried out more effectively when characters similar to one another are used. In other words, the effect of generalization of learning can be expected.

[0042] The learning unit 50 learns the artificial intelligence agents 62 by having two artificial intelligence agents 62 compete against each other, as shown in Fig. 7. That is, in accordance with the rules of the electronic game in question, game information 64 is input to each artificial intelligence agent 62, and the two agents compete against each other, and machine learning of the artificial intelligence agents 62 is carried out according to the result of winning or losing the electronic game.

[0043] In this embodiment, a strength index (strength vector) is input to each artificial intelligence agent 62, and the artificial intelligence agent 62 is trained to achieve a predetermined winning percentage for the relationship between the strength indices. Specifically, as shown in Fig. 8, winning percentages for the relationship between the strength indices are determined in advance and stored in the storage unit 32 as a reward database 66, and the closer the winning percentage in a match between the artificial intelligence agents 62 approaches the winning percentage determined by the relationship between the strength indices input to each artificial intelligence agent 62, the higher the reward is given, and the artificial intelligence agent 62 is trained to obtain a higher reward.

[0044] For example, in the example of the reward database 66 shown in Figure 8, if the strength index (strength vector) input to one artificial intelligence agent 62 is 2 and the strength index (strength vector) input to the other artificial intelligence agent 62 is 3, each artificial intelligence agent 62 is trained to output an electronic game strategy such that in a match between the two artificial intelligence agents 62, one artificial intelligence agent 62 will have a winning rate of 70% and the other artificial intelligence agent 62 will have a winning rate of 30%.

[0045] During learning, the strength index (strength vector) input to each AI agent 62 may be set randomly. Alternatively, the strength index (strength vector) input to one AI agent 62 may be fixed, and the strength index (strength vector) input to the other AI agent 62 may be set randomly.

[0046] This makes it possible to realize an artificial intelligence agent 62 that can adjust its strength in an electronic game according to the reinforcement index (reinforcement vector).

[0047] In this embodiment, the relationship between the strength index (reinforcement vector) and the winning rate is set in the reward database 66, but it is not limited to this and may be anything as long as the winning rate is determined from the strength index (reinforcement vector). For example, the winning rate may be expressed by a function with the strength index (reinforcement vector) as an argument.

[0048] It is also possible to apply reinforcement learning, which is learning for strengthening the artificial intelligence agent 62. As the reinforcement learning, for example, learning may be performed based on the results of a match between the artificial intelligence agent 62 and an NPC (Non Player Character: a rule-based AI that is not the artificial intelligence agent 62) or the results of a match between the artificial intelligence agent 62 and a human player.

[0049] The AI ​​battle processing unit 52 performs processing to battle against a player in an electronic game, as shown in Fig. 9. The AI ​​battle processing unit 52 battles against a player in accordance with the strategy of the electronic game output from the artificial intelligence agent 62 learned by the learning unit 50. A battle against a player is conducted by the player accessing the game server 106 from the player terminal 104.

[0050] When a match between the player and the artificial intelligence agent 62 begins, the AI ​​match processing unit 52 acquires game information 64 related to the electronic game from the player terminal 104 and inputs the game information 64 and a strength index to the artificial intelligence agent 62. The artificial intelligence agent 62 outputs an electronic game strategy corresponding to the input game information 64 and strength index in accordance with a model determined by various parameters (such as the layer structure, the neuron structure in each layer, and the weight coefficient or bias of each neuron) obtained by the above-mentioned learning. The AI ​​match processing unit 52 transmits information for operating an NPC in the electronic game in accordance with the strategy output from the artificial intelligence agent 62 to the player terminal 104. In this way, a match between the player and the artificial intelligence agent 62 is carried out in accordance with the electronic game strategy output from the artificial intelligence agent 62 by inputting game information 64 indicating the state of the electronic game at the time of the match and a strength index to the artificial intelligence agent 62, as shown in FIG. 9 .

[0051] By providing an electronic game using the artificial intelligence agent 62, it is possible for the artificial intelligence agent 62 to output an electronic game strategy with a strength corresponding to the strength index input to the artificial intelligence agent 62. Therefore, the player can play against a computer with a strength corresponding to the strength index.

[0052] Here, it is preferable that the strength index is set according to the player's level in the electronic game. In this embodiment, the strength index is set by the strength index estimator 54 according to the player's level in the electronic game.

[0053] The strength index estimator 54 outputs a strength index for a player according to the player's performance in the electronic game. For example, the strength index estimator 54 may be configured to output a strength index according to the player's win rate in matches against the computer, and the strength index to be input to the artificial intelligence agent 62 may be set according to the strength index output from the strength index estimator 54. Specifically, for example, the strength index estimator 54 may be configured to statistically calculate the win rate in matches against the computer, and output a strength index higher than the strength index currently input to the artificial intelligence agent 62 if the win rate exceeds 50%, and output a strength index lower than the strength index currently input to the artificial intelligence agent 62 if the win rate is lower than 50%.

[0054] The initial value of the strength index input to the artificial intelligence agent 62 may be set randomly or by the player himself.

[0055] Furthermore, the strength index estimator 54 may be configured to output a strength index for a player according to the results of the matches between the players in the electronic game. For example, the strength index estimator 54 may be configured to output a strength index according to the win rate of the player in matches against other players, and the strength index to be input to the artificial intelligence agent 62 may be set according to the strength index output from the strength index estimator 54. Specifically, for example, the strength index estimator 54 may be configured to statistically calculate the win rate in matches between a first player and a second player, and if the win rate of the first player exceeds 50%, output a strength index for the first player that is higher than the strength index already set for the second player, and if the win rate of the first player is lower than 50%, output a strength index for the first player that is lower than the strength index already set for the second player.

[0056] The strength index estimator 54 may also be configured to output a strength index according to the operation history of the player in the electronic data. The player log collection unit 56 acquires a player log, which is historical information on the game progress associated with past plays made by the player in the electronic game, and stores it in the storage unit 32 as a player log DB 68. The strength index estimator 54 may be configured to estimate the strength index of a player by inputting the game information 64 and the player log, and by referring to the electronic game strategy output from the artificial intelligence agent 62 when the same game information 64 is input to the artificial intelligence agent 62, the similarity between the electronic game situation and the player log, the attributes of the player, etc.

[0057] 10, the strength index estimator 54 may be trained to input game information 64, a player log, and a strength index to the strength index estimator 54 and the artificial intelligence agent 62, determine the similarity between the operation content in accordance with the strategy of the electronic game output from the artificial intelligence agent 62 and the operation content actually performed by the player, and estimate and output the player's strength index in accordance with the similarity. Specifically, for example, the strength index estimator 54 may be trained to output, as an estimated value of the player's strength index, a strength index closer to the strength index currently input to the strength index estimator 54, the higher the similarity between the operation content in accordance with the strategy of the electronic game output from the artificial intelligence agent 62 and the operation content actually performed by the player.

[0058] Specifically, the strength index estimator 54 may be configured with a model including a multilayer neural network, and the model may be subjected to machine learning. The multilayer neural network is defined by various parameters (such as the layer structure, the neuron structure in each layer, and the weight or bias of each neuron). Therefore, the memory unit 32 stores, as the strength index estimation model 70, various parameters related to the network that constitutes the model (such as the layer structure, the neuron structure in each layer, and the weight coefficient or bias of each neuron).

[0059] However, the learning method of the strength index estimator 54 is not limited to this, and any method may be used as long as it can estimate a strength index for a player in accordance with an input of a player log. For example, an existing method such as a multi-armed bandit algorithm may be applied to train the strength index estimator 54 so that it can estimate a strength index for a player in accordance with an input of a player log.

[0060] 9, when playing an electronic game using an artificial intelligence agent 62, it is preferable to set the strength index input to the artificial intelligence agent 62 to the strength index of the player estimated by the strength index estimator 54. In this case, it is preferable to set the maximum value of the strength index to be set for the artificial intelligence agent 62 to the maximum value of the strength index used when training the artificial intelligence agent 62.

[0061] In this way, in a match between a player and an artificial intelligence agent 62, by dynamically adjusting the strength index input to the artificial intelligence agent 62 in accordance with the strength index estimated for the player, it is possible to quantitatively optimize the strength of the artificial intelligence agent 62 without manually adjusting the strength by a planner, the player himself, or the like.

[0062] Although the embodiment of the present invention has been described above, the present invention is not limited to the above embodiment, and various modifications are possible without departing from the spirit of the present invention. [Explanation of symbols]

[0063] 10 processing unit, 12 memory unit, 14 input unit, 16 output unit, 18 communication unit, 20 processing unit, 22 memory unit, 24 input unit, 26 output unit, 28 communication unit, 30 processing unit, 32 memory unit, 34 input unit, 36 output unit, 38 communication unit, 50 learning unit, 52 AI battle processing unit, 54 strength index estimator, 56 player log collection unit, 62 artificial intelligence agent, 64 game information, 66 reward database, 68 player log DB, 70 strength index estimation model, 100 electronic game providing system, 102 planner terminal, 104 player terminal, 106 game server, 108 information and communication network, 200 board, 202 hand pieces.

Claims

1. an artificial intelligence agent that is trained to use at least game information including information related to game progress in an electronic game and a strength index indicating strength in the electronic game to output a play strategy in the electronic game with a strength corresponding to the strength index; The electronic game provision system is characterized in that the artificial intelligence agent uses a reward database that associates the strength index with the winning rate in the electronic game, and gives higher rewards to the artificial intelligence agents in a match between themselves as the winning rate is closer to the winning rate determined by the relationship between the strength indices input to each of the artificial intelligence agents, thereby training the artificial intelligence agent to obtain higher rewards.

2. 2. The electronic game providing system according to claim 1, In a battle between said AI agents, the strength index input to each AI agent is set randomly.

3. 3. The electronic game providing system according to claim 1, an estimator for estimating a player strength index indicating a strength of a player when playing the electronic game; An electronic game providing system, wherein when the electronic game is played using the artificial intelligence agent, a strength index input to the artificial intelligence agent is set in accordance with the player strength index.

4. 4. The electronic game providing system according to claim 3, The electronic game providing system is characterized in that the estimator estimates the player strength index based on a progress history of the player's play of the electronic game.

5. 4. The electronic game providing system according to claim 3, The electronic game providing system, wherein the estimator estimates the player strength index based on the results of the player's play of the electronic game.

6. 6. The electronic game providing system according to claim 5, The electronic game providing system is characterized in that the estimator estimates the player strength index based on a statistical winning rate of the player's play of the electronic game.

7. The electronic game providing system according to any one of claims 4 to 6, The electronic game providing system is characterized in that the estimator includes an artificial intelligence estimator that has been trained to input historical information from when a player played the electronic game and output the player's strength in the electronic game.

8. Computer, using at least game information including information related to game progress in the electronic game and a strength index indicating strength in the electronic game, the agent functions as an artificial intelligence agent that has been trained to output a play strategy in the electronic game with a strength corresponding to the strength index; The artificial intelligence agent uses a reward database that associates the strength index with the winning rate in the electronic game, and in a match between the artificial intelligence agents, the artificial intelligence agent is given a higher reward the closer the winning rate determined by the relationship between the strength indexes input to each artificial intelligence agent, thereby training the artificial intelligence agent to obtain a higher reward.

9. Computer, using at least game information including information related to game progress in the electronic game and a strength index indicating strength in the electronic game, the agent functions as an artificial intelligence agent that has been trained to output a play strategy in the electronic game with a strength corresponding to the strength index; A method for providing an electronic game in which the artificial intelligence agent uses a reward database that associates the strength index with the winning rate in the electronic game, and gives a higher reward the closer the winning rate in a match between the artificial intelligence agents is to the winning rate determined by the relationship between the strength indexes input to each artificial intelligence agent, thereby training the artificial intelligence agent to obtain a higher reward.

10. Computer, using at least game information including information related to game progress in the electronic game and a strength index indicating strength in the electronic game, the artificial intelligence agent is trained to output a play strategy in the electronic game with strength corresponding to the strength index; The method for providing an electronic game is characterized in that the artificial intelligence agent uses a reward database that associates the strength index with the winning rate in the electronic game, and gives higher rewards to the artificial intelligence agents in a match between themselves as the winning rate is closer to the winning rate determined by the relationship between the strength indices input to each of the artificial intelligence agents, thereby training the artificial intelligence agent to obtain higher rewards.

Images