Autonomous Agents and Multiagent Systems. Best and Visionary Papers : AAMAS 2023 Workshops, London, UK, May 29 –June 2, 2023, Revised Selected Papers / edited by Francesco Amigoni, Arunesh Sinha.

Format:

Book

Author/Creator:

Amigoni, Francesco.

Contributor:

Sinha, Arunesh.

Series:

Lecture Notes in Artificial Intelligence, 2945-9141 ; 14456

Language:

English

Subjects (All):

Artificial intelligence.

Computer engineering.

Computer networks.

Software engineering.

Social sciences--Data processing.

Social sciences.

Numerical analysis.

Artificial Intelligence.

Computer Engineering and Networks.

Software Engineering.

Computer Application in Social and Behavioral Sciences.

Numerical Analysis.

Local Subjects:

Artificial Intelligence.

Computer Engineering and Networks.

Software Engineering.

Computer Application in Social and Behavioral Sciences.

Numerical Analysis.

Physical Description:

1 online resource (236 pages)

Edition:

1st ed. 2024.

Place of Publication:

Cham : Springer Nature Switzerland : Imprint: Springer, 2024.

Summary:

This book contains visionary and best papers from the workshops held at the International Conference on Autonomous Agents and Multiagent Systems, AAMAS 2023, held in London, UK, during May 29–June 2, 2023. The 12 regular papers, 5 best papers and 7 visionary papers, presented were carefully reviewed and selected from a total of more than 110 contributions to the workshops. They focus on emerging topics and new trends in the area of autonomous agents and multiagent systems and stem from the following workshops: - Workshop on Autonomous Robots and Multirobot Systems (ARMS) - Workshop on Adaptive and Learning Agents (ALA) - Workshop on Interdisciplinary Design of Emotion Sensitive Agents (IDEA) - Workshop on Rebellion and Disobedience in Artificial Intelligence (RaD-AI) - Workshop on Neuro-symbolic AI for Agent and Multi-Agent Systems (NeSyMAS) - Workshop on Multiagent Sequential Decision Making under Uncertainty (MSDM) - Workshop on Citizen-Centric Multi-Agent Systems (C-MAS) .

Contents:

Intro

Preface

Organization

Contents

Best Papers

Fair Deep Reinforcement Learning with Generalized Gini Welfare Functions

1 Introduction

2 Related Work

2.1 Fairness in AI and ML

2.2 Fairness in RL

2.3 State Augmentation

3 Background

3.1 Markov Decision Process and RL

3.2 Multiobjective Markov Decision Process

3.3 Deep RL

3.4 Fairness

4 Fair Optimization with Preferential Treatment

4.1 Generalized GGF

4.2 Problem Statement

4.3 Difficulties

5 Proposed Algorithms

5.1 Value-Based Methods

5.2 Policy Gradient Methods

6 Experimental Results

6.1 Species Conservation

6.2 Traffic Light Control

6.3 Data Center

7 Conclusion

References

The Value of Knowledge: Joining Reward and Epistemic Certainty Optimisation for Anxiety-Sensitive Planning

2 Background

2.1 Anxiety

2.2 Partially Observable Markov Decision Processes

2.3 Epistemic Partially Observable Markov Decision Processes

3 R-POMDP Formalism

4 Anxiety Models

5 Anxiety-Sensitive Planning Algorithm

6 Experiments

6.1 The Bridge Inspection Problem

6.2 Setup

6.3 Results

7 Discussion and Further Works

8 Concluding Remarks

Learning Reward Machines in Cooperative Multi-agent Tasks

2.1 Reinforcement Learning

2.2 Reward Machines

3 Learning Reward Machine in Cooperative Multi-agent Tasks

3.1 Learn a Global Reward Machine

3.2 Learn Individual Reward Machine

4 Experiments

4.1 ThreeButtons Task

4.2 Rendezvous Task

5 Related Work

6 Conclusion and Future Work

Discrete-Choice Multi-agent Optimization: Decentralized Hard Constraint Satisfaction for Smart Cities

2 Comparison to Related Work

3 Hard Constraint Satisfaction Model

3.1 Optimization Problem.

3.2 A Heuristic for Satisfying Hard Constraints

4 Hard Constraints Implementation

5 Experimental Evaluation

5.1 Smart City Application Scenarios

5.2 Hard Constraint Satisfaction Works

5.3 Behavior Shift Can Mitigate Hard Constraints

Deliberation and Voting in Approval-Based Multi-winner Elections

2 Preliminaries

3 Deliberation

4 Experimental Evaluation

5 Conclusion

A Preliminaries and Definitions

A.1 Properties

A.2 Multi-winner Voting Rules

B Deliberation Models

B.1 Voting Population: Preferences and Utilities

B.2 The Deliberation Process

C Opinion Dynamics Models

D Iterative Golfer

E Further Details About the Experimental Setup

F Further Results from Sect.4

F.1 EJR and PJR Satisfaction

F.2 Inter-group Ballot Disagreement

F.3 Voter Satisfaction

G Iterative Deliberation with CC

Visionary Papers

Plan Generation via Behavior Trees Obtained from Goal-Oriented LTLf Formulas

3 Behavior Tree and LTLf Semantics

3.1 Behavior Tree Semantics

3.2 LTLf Semantics

4 Constructing a Behavior Tree from an LTLf Task Formula

4.1 Task Formula and Task Behavior Tree

4.2 BT Success Implies Task Formula Satisfied

4.3 Empirical Verification

5 BT for LTLf Task-Based Missions

5.1 Constructing a Mission Behavior

5.2 BT Success Implies Formula Satisfied

6 Planner-Goal Alignment Example

6.1 Problem Formulation

6.2 Action-Node Policies via Policy Iteration

6.3 Action-Node Policies with BT Feedback

7 Fetch Robot Example

7.1 Problem Specification

7.2 Experiment Design

7.3 Results

8 Summary and Future Work

Discovery and Analysis of Rare High-Impact Failure Modes Using Adversarial RL-Informed Sampling

1 Introduction.

2 Adaptive Stress Testing

2.1 Limitations

3 Failure Trace Resampling

3.1 Error Estimation

3.2 Multimodal Failure Trace Resampling

4 Failure Policies

5 Failure Policy Resampling

7 Discussion and Future Work

8 Conclusion

Anxiety Among Migrants - Questions for Agent Simulation

2 Modelling

2.1 Modelling Emotions

2.2 Classical Theory of Emotions

2.3 Affective Neuroscience

2.4 Computer Science

3 Our Currrent Agent Model

3.1 The Multi Queue Model Approach

4 Technical Details

4.1 Adding Anxiety to the Agent Model

4.2 Applying the Computational Models

Should My Agent Lie for Me? Public Moral Perspectives on Deceptive AI

2 Background and Related Work

3 Methods

4 Results

4.1 Moral Permissibility of Deception

4.2 Trust in Agent

4.3 Willingness to Buy

4.4 Responsibility Assignment

5 Discussion

6 Limitations and Future Work

7 Towards a Normative Framework for Deceptive AI

7.1 Accountable, Responsible, and Transparent Deceptive AI

7.2 The Ethically Aligned Deceptive Machine Argument

7.3 The Scientific Discovery Argument in AI

7.4 Towards Trustworthy Deceptive AI

8 Conclusions

Neuro-Symbolic AI + Agent Systems: A First Reflection on Trends, Opportunities and Challenges

2 Key Considerations

3 Should We Care that Neuro-Symbolic AI Is a ``Broad Church'' Covering Many varieties?

4 How Can a Balance Between Reasoning And learning Be Achieved in the Context of Agents / robots?

5 In Practice: How Much Logic Is Really needed? How Much Learning Is Really needed?

6 What Can Agent Researchers Learn From Neuro-Symbolic AI?.

7 Are There application Areas Where We Clearly Need More ``neuro'' and Are There others Where We Clearly Need More ``symbolic''?

8 Which Industry Is ``ripe'' For Neuro-Symbolic AI?

9 Is the Answer Large Language Models?

10 What Are the Barriers to Using Neuro-Symbolic AI? What Issue Needs to Be Fixed next?

11 Current Trends in Neuro-Symbolic AI

12 Conclusions

Novelty Accommodating Multi-agent Planning in High Fidelity Simulated Open World

3 Problem Setup

4 Planning Domain Formalization

5 Implementing Novelty Accommodating Agents

5.1 High Level Architecture

5.2 PDDL+ and Execution Engine

5.3 Novelty Injection

5.4 Consistency Checking

5.5 Meta Model Repair

6 Experiment Setup

6.1 Novelties

7 Results

8 Conclusion and Future Work

Towards Reducing School Segregation by Intervening on Transportation Networks

2 Methods

2.1 Environment: Citizens, Transportation, and Schools

2.2 Preference Model

2.3 Allocation Method

2.4 Allocation Evaluation

2.5 Intervention Model

3 Experimental Setup

4 Preliminary Results

5 Conclusion and Future Work

A Intervention Methods

A.1 Greedy Centrality Optimization

A.2 Group-Based Centrality

B Simulation Environments

B.1 Simulation Parameters

Author Index.

Notes:

Description based on publisher supplied metadata and other sources.

ISBN:

9783031562556

3031562550

OCLC:

1428319912