Towards Equilibrium Transfer in Markov Games 胡裕靖 2013-9-9.

Slides:



Advertisements
Similar presentations
Alina Pommeranz, MSc in Interactive System Engineering supervised by Dr. ir. Pascal Wiggers and Prof. Dr. Catholijn M. Jonker.
Advertisements

Continuation Methods for Structured Games Ben Blum Christian Shelton Daphne Koller Stanford University.
Game Theoretical Insights in Strategic Patrolling: Model and Analysis Nicola Gatti – DEI, Politecnico di Milano, Piazza Leonardo.
Learning in Multi-agent System
LAUREN FRATAMICO Common knowledge of rationality and backward induction.
Meta-Level Control in Multi-Agent Systems Anita Raja and Victor Lesser Department of Computer Science University of Massachusetts Amherst, MA
Extraction and Transfer of Knowledge in Reinforcement Learning A.LAZARIC Inria “30 minutes de Science” Seminars SequeL Inria Lille – Nord Europe December.
Game Theoretic Aspect in Human Computation Presenter: Chien-Ju Ho
Background Reinforcement Learning (RL) agents learn to do tasks by iteratively performing actions in the world and using resulting experiences to decide.
Introduction to Hierarchical Reinforcement Learning Jervis Pinto Slides adapted from Ron Parr (From ICML 2005 Rich Representations for Reinforcement Learning.
David Wingate Reinforcement Learning for Complex System Management.
Satisfaction Equilibrium Stéphane Ross. Canadian AI / 21 Problem In real life multiagent systems :  Agents generally do not know the preferences.
Non-cooperative Game Theory Notes by Alberto Bressan.
Game-Theoretic Approaches to Multi-Agent Systems Bernhard Nebel.
Gabriel Tsang Supervisor: Jian Yang.  Initial Problem  Related Work  Approach  Outcome  Conclusion  Future Work 2.
Supporting Design Managing complexity of designing Expressing ideas Testing ideas Quality assurance.
Lectures in Microeconomics-Charles W. Upton Game Theory.
INSTITUTO DE SISTEMAS E ROBÓTICA Minimax Value Iteration Applied to Robotic Soccer Gonçalo Neto Institute for Systems and Robotics Instituto Superior Técnico.
Outline MDP (brief) –Background –Learning MDP Q learning Game theory (brief) –Background Markov games (2-player) –Background –Learning Markov games Littman’s.
Research Related to Real-Time Strategy Games Robert Holte November 8, 2002.
Cooperative Q-Learning Lars Blackmore and Steve Block Expertness Based Cooperative Q-learning Ahmadabadi, M.N.; Asadpour, M IEEE Transactions on Systems,
Hierarchical Reinforcement Learning Ersin Basaran 19/03/2005.
Better automated abstraction techniques for imperfect information games, with application to Texas Hold’em poker * Andrew Gilpin and Tuomas Sandholm, CMU,
Reinforcement Learning (1)
1 Issues on the border of economics and computation נושאים בגבול כלכלה וחישוב Congestion Games, Potential Games and Price of Anarchy Liad Blumrosen ©
Simple search methods for finding a Nash equilibrium Ryan Porter, Eugene Nudelman, and Yoav Shoham Games and Economic Behavior, Vol. 63, Issue 2. pp ,
Strategic Modeling of Information Sharing among Data Privacy Attackers Quang Duong, Kristen LeFevre, and Michael Wellman University of Michigan Presented.
BACKGROUND KNOWLEDGE IN ONTOLOGY MATCHING Pavel Shvaiko joint work with Fausto Giunchiglia and Mikalai Yatskevich INFINT 2007 Bertinoro Workshop on Information.
Learning in Multiagent systems
General Polynomial Time Algorithm for Near-Optimal Reinforcement Learning Duke University Machine Learning Group Discussion Leader: Kai Ni June 17, 2005.
Cognitive Apprenticeship “Mastering knowledge” CLICK TO START.
1 ECE-517 Reinforcement Learning in Artificial Intelligence Lecture 7: Finite Horizon MDPs, Dynamic Programming Dr. Itamar Arel College of Engineering.
Transfer Learning with Applications to Text Classification Jing Peng Computer Science Department.
Reinforcement Learning Presentation Markov Games as a Framework for Multi-agent Reinforcement Learning Mike L. Littman Jinzhong Niu March 30, 2004.
Leveraging Human Knowledge for Machine Learning Curriculum Design Matthew E. Taylor teamcore.usc.edu/taylorm.
Reinforcement Learning Ata Kaban School of Computer Science University of Birmingham.
Learning the Structure of Related Tasks Presented by Lihan He Machine Learning Reading Group Duke University 02/03/2006 A. Niculescu-Mizil, R. Caruana.
Exploring the role of visualization and engagement in Computer Science Education Naps, T., & etc. (2003). Exploring the role of visualization and engagement.
Cooperative Q-Learning Lars Blackmore and Steve Block Multi-Agent Reinforcement Learning: Independent vs. Cooperative Agents Tan, M Proceedings of the.
Regularization and Feature Selection in Least-Squares Temporal Difference Learning J. Zico Kolter and Andrew Y. Ng Computer Science Department Stanford.
1 Multiagent Teamwork: Analyzing the Optimality and Complexity of Key Theories and Models David V. Pynadath and Milind Tambe Information Sciences Institute.
CSCI N100 Principles of Computing Basic Problem-Solving.
A Theoretical Analysis of Multi-Agent Patrolling Strategies Patrolling = moving through a territory « visiting » areas The patrolling problem = given a.
1 Ann Nowé Nature inspired agents to handle interaction in IT systems Ann Nowé Computational modeling Lab Vrije Universiteit Brussel.
Dueling Estimates: Closing Thoughts Jennifer H. Madans National Center for Health Statistics.
The ideals reality of science The pursuit of verifiable answers highly cited papers for your c.v. The validation of our results by reproduction convincing.
History of Management Thought
Information Technology Michael Brand Joint work with David L. Dowe 8 February, 2016 Information Technology.
Modeling & Simulation of Dynamic Systems (MSDS)
Generalized Point Based Value Iteration for Interactive POMDPs Prashant Doshi Dept. of Computer Science and AI Institute University of Georgia
ECO290E: Game Theory Lecture 8 Games in Extensive-Form.
Web-Mining Agents: Transfer Learning TrAdaBoost R. Möller Institute of Information Systems University of Lübeck.
1 Nash Demand Game Nash Program (non cooperative games) Demand Game S Topics 3.
On the Difficulty of Achieving Equilibrium in Interactive POMDPs Prashant Doshi Dept. of Computer Science University of Georgia Athens, GA Twenty.
Matthew E. Taylor 1 Autonomous Inter-Task Transfer in Reinforcement Learning Domains Matthew E. Taylor Learning Agents Research Group Department of Computer.
Reinforcement Learning Guest Lecturer: Chengxiang Zhai Machine Learning December 6, 2001.
CSCI N207 Data Analysis Using a Spreadsheet. Course Goals This is a technical course in which data analysis is the main emphasis and spreadsheet is the.
ECO290E: Game Theory Lecture 10 Examples of Dynamic Games.
Transfer and Multi-Task Learning in Reinforcement Learning Alessandro LAZARIC “Machine Learning with Interdependent and Non-identically Distributed Data”
Transfer Learning and Intelligence: an Argument and Approach Matthew E. Taylor Joint work with: Gregory Kuhlmann and Peter Stone Learning Agents Research.
Game theory basics A Game describes situations of strategic interaction, where the payoff for one agent depends on its own actions as well as on the actions.
Dynamics of Learning & Distributed Adaptation
Object-Oriented Software Engineering Using UML, Patterns, and Java,
Transferring Instances for Model-Based Reinforcement Learning
Structured Models for Multi-Agent Interactions
CASE − Cognitive Agents for Social Environments
Multiagent Systems Repeated Games © Manfred Huber 2018.
Chapter 7: Eligibility Traces
Collaboration in Repeated Games
Normal Form (Matrix) Games
Presentation transcript:

Towards Equilibrium Transfer in Markov Games 胡裕靖

Outline  Background  Preliminary Ideas  Some Results

Background

Multi-agent Reinforcement Learning Single-agent RL: Mountain Car Path finding RL in multi-agent tasks Robot Soccer IKEA furniture robot

Markov Games Agent take joint actions from one agent to more than one

Equilibrium-based MARL Some equilibrium solution concepts in game theory can be adopted

Our Previous Work  Equilibrium-based MARL: Multi-agent reinforcement learning with meta equilibrium [] Multi-agent reinforcement learning by negotiation with unshared value functions [] Focusing on combining MARL with equilibrium solution concepts  Problematic issues: Equilibrium computing is complicated and time consuming A new complexity class: TFNP! [] For tasks with many agents, equilibrium-based MARL algorithms may take too much time How to accelerate the learning process of equilibrium-based MARL?

Transfer Learning in RL Matthew E Taylor, Peter Stone. Transfer learning for reinforcement learning domains. Journal of Machine Learning Research, instance/policy/value function/model/… Alessandra Lazaric. Transfer in reinforcement learning: a framework and a survey. Reinforcement Learning, Springer, accelerate Reuse learnt knowledge

Transfer Learning in Markov Games? instance/policy/value function/model/… …… Why not transfer between these normal-form games within a Markov game? Inter-task transfer Inner-task transfer

Inner-task Transfer  Transfer equilibrium between similar normal-form games during learning in a Markov game: Reuse the computed equilibria in previous games Reducing learning time  Key problems: Which games are similar? For example: the games occur on different visits of a state How to transfer equilibrium?

Preliminary Ideas

Game Similarity  Games with the same action space?  Games with different action space?  Similarity payoff distance?  Equilibrium-based similarity or equilibrium-independent similarity? Drew Fudenberg and David M. Kreps. A theory of learning, experimentation and equilibrium in games

Game Similarity Equilibrium-based similarity Equilibrium transfer Find equilibria of two games and compute the similarity Transfer seems senseless! Weird Cycle

Our Idea Transfer equilibrium between games which are thought to be similar. Evaluate how much the loss brought by equilibrium transfer is. Transfer is acceptable when there is a little loss. The two games are different only in one item.

Problem Definition Approximate Nash equilibrium

Problem Definition

A Naïve Transfer Method Direct Transfer

A Naïve Transfer Method

Some Results

Future Work  Some problems: Other transfer methods? Only Nash equilibrium? Equilibrium finding algorithms  Transfer between games with different action space  Transfer between games with different agent numbers  Game abstraction

Thanks!