Multi-agent game playing method based on spiking neural network

A pulse neural network and multi-agent technology, applied in the field of neural network and artificial intelligence, can solve problems such as inability to make real-time decisions, inability to perform network self-growth and self-organization, and inability to learn and reason, achieving outstanding learning ability, Save computing power and time, and achieve high degree of intelligence

Pending Publication Date: 2022-03-08
杨旭
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

These practical problems require more and more bionic agents. Whether a group of agents can work like humans in a dynamic and uncertain environment is the key to solving these problems. However, the current multi-agent technology is still Difficulty meeting challenges in complex situations
Deep learning and deep reinforcement learning use their own powerful information processing capabilities to show advantages in multi-agent games, but they still face challenges, mainly manifested in the inability to self-grow and self-organize the network, so in the face of uncertain factors Real-time decision-making is not possible, and the ability to learn and reason is not available

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Multi-agent game playing method based on spiking neural network
  • Multi-agent game playing method based on spiking neural network
  • Multi-agent game playing method based on spiking neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0109] Taking the multi-agent game under the condition of non-perfect information based on the pulse neural network - Doudizhu playing cards as an example, the steps and effects of the present invention are described in detail.

[0110] In this example, there are three players playing a game of gambling with a deck of cards (54 cards in total). In a game, it is divided into two parties, one player is the "landlord" and is one party, and the remaining two players are farmers and are the other party. The game stipulates that the first party to deal out all the cards in its hand wins. When a player plays a card, the card played must be the player's own hand card, and must be consistent with the rules of the card type played by the previous player, and the value of the card must be greater than that of the previous player.

[0111] In a round of the game, according to the learning object, one player to be trained is regarded as the ontology, and the remaining two players are rega...

Embodiment 2

[0123] Taking the multi-agent game under imperfect information conditions based on the pulse neural network - Texas Hold'em as an example, the steps and effects of the present invention are specifically described.

[0124] In this example, there are 2-10 players in total, and a deck of 52 cards with queens and queens removed is used for gaming. At the beginning of the game, each player will be issued 2 "bottom cards" (only individuals can see), and three public cards, 1 and 1, will be issued on the table three times in succession. After four rounds of calling, raising After betting, discarding and other betting operations, enter the showdown stage, choose the largest 5-card combination from your 2 hole cards and 5 public cards, and decide the winner according to the rules of card size. The winner takes all the chips .

[0125] In a round of the game, according to the learning object, one player to be trained is regarded as the ontology, and the remaining players are regarded ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

A multi-agent game method based on a spiking neural network comprises the following steps: dividing a multi-agent area in an environment into a body and other agents, establishing input layers of the other agents except the body, and generating and exciting input layer neurons; establishing a rule layer corresponding to input layers of other agents, establishing an input layer of the ontology, and generating and exciting neurons of the input layer; establishing a rule layer corresponding to the ontology input layer, establishing a decision-making layer of the multi-agent game, and generating intra-layer synaptic connection of the decision-making layer according to the excitation condition of neurons in the decision-making layer; and establishing an output layer of the multi-agent game, establishing one-to-one connection between the neurons of the decision layer and the neurons of the output layer, and obtaining a final output signal by adjusting the connection weight of the decision layer and the output layer. The method not only has learning reasoning ability, but also greatly reduces operation requirements, and is suitable for multi-agent gaming in various environments.

Description

technical field [0001] The invention belongs to the technical field of artificial intelligence and neural network, in particular to a method for multi-agent game based on impulse neural network. Background technique [0002] In the real world, there are a large number of complex dynamic decision-making problems, such as road traffic system, economic forecasting, military decision-making, etc. These practical problems require more and more bionic agents. Whether a group of agents can work like humans in a dynamic and uncertain environment is the key to solving these problems. However, the current multi-agent technology is still Difficulty meeting challenges in complex situations. Deep learning and deep reinforcement learning use their own powerful information processing capabilities to show advantages in multi-agent games, but they still face challenges, mainly manifested in the inability to self-grow and self-organize the network, so in the face of uncertain factors Real-t...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06N3/04G06N3/06G06N3/08
CPCG06N3/061G06N3/08G06N3/045
Inventor 董丽亚杨旭晏子华林深吉梦瑶郑文浩赵晋锋张志松王麒淋
Owner 杨旭
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap