New Zealand

Dennis Barrios-Aranibar and Luiz M. G. Gon&#231;alves (January 1st 2009). Influence Value Q-Learning: A Reinforcement Learning Algorithm for Multi Agent Systems, Theory and Novel Applications of Machine Learning, Meng Joo Er and Yi Zhou, IntechOpen, DOI: 10.5772/6675. 

Collective learning can be greatly enhanced when agents effectively exchange knowledge with their peers. In particular, recent work studying agents that learn to teach other teammates has demonstrated that action advising accelerates team-wide learning. However, the prior work has simplified the learning of advising policies by using simple function approximations and only considered advising with primitive (low-level) actions, limiting the scalability of learning and teaching to complex domains. This paper introduces a novel learning-to-teach framework, called hierarchical multiagent teaching (HMAT), that improves scalability to complex environments by using the deep representation for student policies and by advising with more expressive extended action sequences over multiple levels of temporal abstraction. Our empirical evaluations demonstrate that HMAT improves team-wide learning progress in large, complex domains where previous approaches fail. HMAT also learns teaching policies that can effectively transfer knowledge to different teammates with knowledge of different tasks, even when the teammates have heterogeneous action spaces.

AAMAS 2020

Learning Hierarchical Teaching Policies for Cooperative Agents

deep multiagent reinforcement learning

learning to teach

inter-agent knowledge transfer

The 9th Brazilian Conference on Intelligent Systems (BRACIS), the 16h Encontro Nacional de Inteligência Artificial e Computacional (ENIAC), and the 12th Concurso de Teses e Dissertações em Inteligência Artificial e Computacional (CTDIAC) was held from 20 to 23 October, 2020.

BRACIS originated from the combination of the two most important scientific events in Brazil in Artificial Intelligence (AI) and Computational Intelligence (CI): the Brazilian Symposium on Artificial Intelligence - SBIA (22 editions), and the Brazilian Symposium on Neural Networks - SBRN (13 editions). ENIAC is a national forum for researchers, practitioners, educators and students to present and discuss innovations, trends, experiences, developments and work in progress in the fields of Artificial and Computational Intelligence. CTDIAC is a satellite event whose goals are to disseminate and award the best dissertations and theses (MSc and PhDs concluded in Brazilian universities from June 2018 to May 2020) in Artificial and Computational Intelligence areas.

In this year, the Symposium on Knowledge Discovery, Mining and Learning (KDMiLe) will be held in conjunction with BRACIS 2020. This sysposium aims at integrating researchers, practitioners, developers, students and users to present their research results, to discuss ideas, and to exchange techniques, tools, and practical experiences – related to the Data Mining and Machine Learning areas.

BRACIS 2020

technical paper

AAMAS is the leading scientific conference for research in autonomous agents and multi-agent systems. The AAMAS conference series was initiated in 2002 as the merging of three respected scientific meetings: the International Conference on Multi-Agent Systems (ICMAS), the International Workshop on Agent Theories, Architectures, and Languages (ATAL), and the International Conference on Autonomous Agents (AA). The aim of the joint conference is to provide a single, high-profile, internationally-respected archival forum for scientific research in the theory and practice of autonomous agents and multi-agent systems.

Browse keynotes, discussions, panels and over 300 presentations.


AAMAS is the leading scientific conference for research in autonomous agents and multi-agent systems. The AAMAS conference series was initiated in 2002 as the merging of three respected scientific meetings: the International Conference on Multi-Agent Systems (ICMAS), the International Workshop on Agent Theories, Architectures, and Languages (ATAL), and the International Conference on Autonomous Agents (AA). 

In classical elections, voters only submit their ballot once, whereas, in iterative voting, the ballots may be changed iteratively. Following the work by Wilczynski [2019], we consider the case where a social network represents an underlying structure between the voters, meaning that each voter can see her neighbors’ ballots. In addition, there is a polling agency, which publicly announces the result for the initial vote. This paper investigates the manipulative power of the polling agency. Previously, Wilczynski [2019] studied constructive manipulation for the plurality rule. We introduce destructive manipulation and extend the study to the veto rule. Several restricted variants are considered with respect to their parameterized complexity. The theoretical results are complemented by experiments
using different heuristics.

Manipulation of Opinion Polls to Influence Iterative Elections

Off Policy Deep Reinforcement Learning with Analogous Disentangled Exploration

Reinforcement learning (RL), like any on-line learning method, inevitably faces the exploration-exploitation dilemma. When a learning algorithm requires as few data samples as possible, it is called sample efficient. The design of sample-efficient algorithms is an important area of research. Interestingly, all currently known provably efficient model-free RL algorithms utilize the same well-known principle of optimism in the face of uncertainty. We unite these existing algorithms into a single general model-free optimistic RL framework. We show how this facilitates the design of new optimistic model-free RL algorithms by simplifying the analysis of their efficiency. Finally, we propose one such new algorithm and demonstrate its performance in an experimental study.

Generalized Optimistic Q-Learning with Provable Efficiency

A patrol of robot teams, where the robots are required to repeatedly visit a target area, is a useful tool in detecting an adversary trying to penetrate. In this work we examine the Closed Perimeter Patrol problem, in which the robots travel along a closed perimeter and the adversary is aware of the robots' patrol policy. The goal is to maximize the probability of penetration detection. Previous work dealt with symmetric tracks, in which all parts of the track have similar properties, and suggested non-deterministic patrol schemes, characterized by a uniform policy along the entire area. We consider more realistic scenarios of asymmetric tracks, with various parts of the track having different properties, and suggest a patrol policy with a non-uniform policy along different points of the track. We compare the achievements of both models and show the advantage of the non-uniform model. We further explore methods to efficiently calculate the attributes needed to maximize the probability of penetration detection and compare their implementation in various scenarios.


Non-Uniform Policies for Multi-Robot Asymmetric Perimeter Patrol in Adversarial Domains

Multi-Agent Path Finding (MAPF) plays an important role in many real-life applications where autonomous agents must coordinate to reach their goals without collisions. MAPF problems often take place in structured environments that are usually assumed to be static and known in advance. In this paper, we introduce C-MAPF, i.e., MAPF in Configurable environments, a novel variant of the MAPF problem in which the environment is configurable, namely its structure and topology can be controlled within some given constraints. Consider, for instance, a warehouse logistics application: the environment can be changed (at least to some degree) by the managers of the warehouse, for example by re-arranging the positions of the shelves or by removing or adding temporary walls. We study the properties of the C-MAPF problem and we devise two algorithms for solving it, both based on Conflict-Based Search (CBS), a state-of-the-art MAPF algorithm. First, we present Parallel CBS (P-CBS), that searches for a solution by simultaneously considering all the possible configurations of the environment. We then present Abstract CBS (A-CBS), an extended version of the CBS algorithm that solves C-MAPF problems by introducing a new type of conflict on the allowable configurations of the environment. We prove that our solvers are both complete and optimal and we experimentally assess their performance in different settings.


Multi-Agent Path Finding in Configurable Environments

In many settings where multiple agents interact, the optimal choices for each agent depend heavily on the choices of the others.
These coupled interactions are well-described by a general-sum differential game, in which players have differing objectives, the state evolves in continuous time, and optimal play may be characterized by one of many equilibrium concepts, e.g., a Nash equilibrium.
Often, problems admit multiple equilibria.
From the perspective of a single agent in such a game, this multiplicity of solutions can introduce uncertainty about how other agents will behave.
This paper proposes a general framework for resolving ambiguity between equilibria by reasoning about the equilibrium other agents are aiming for.
We demonstrate this framework in simulations of a multi-player human-robot navigation problem that yields two main conclusions:
First, by inferring which equilibrium humans are operating at, the robot is able to predict trajectories more accurately, and second, by discovering and aligning itself to this equilibrium the robot is able to reduce the cost for all players.


Inference-Based Strategy Alignment for General-Sum Differential Games

Multiagent teams have been shown to be effective in many domains that require coordination among team members. However, finding valuable joint-actions becomes increasingly difficult in tightly-coupled domains where each agent's performance depends on the actions of many other agents. Reward shaping partially addresses this challenge by deriving more "tuned" rewards to provide agents with additional feedback, but this approach still relies on agents randomly discovering suitable joint-actions. In this work, we introduce Counterfactual Agent Suggestions (CAS) as a method for injecting knowledge into an agent's learning process within the confines of existing reward structures. We show that CAS enables agent teams to converge towards desired behaviors more reliably. We also show that improvement in team performance in the presence of suggestions extends to large teams and tightly-coupled domains.


The Power of Suggestion

Reinforcement learning systems require good representations to work well. For decades practical success in reinforcement learning was limited to small domains. Deep reinforcement learning systems, on the other hand, are scalable, not dependent on domain specific prior knowledge and have been successfully used to play Atari, in 3D navigation from pixels, and to control high degree of freedom robots. Unfortunately, the performance of deep reinforcement learning systems is sensitive to hyper-parameter settings and architecture choices. Even well tuned systems exhibit significant instability both within a trial and across experiment replications. In practice, significant expertise and trial and error are usually required to achieve good performance. One potential source of the problem is known as catastrophic interference: when later training decreases performance by overriding previous learning. Interestingly, the powerful generalization that makes Neural Networks (NN) so effective in batch supervised learning might explain the challenges when applying them in reinforcement learning tasks. In this paper, we explore how online NN training and interference interact in reinforcement learning. We find that simply re-mapping the input observations to a high-dimensional space improves learning speed and parameter sensitivity. We also show this preprocessing reduces interference in prediction tasks. More practically, we provide a simple approach to NN training that is easy to implement, and requires little additional computation. We demonstrate that our approach improves performance in both prediction and control with an extensive batch of experiments in classic control domains.

Improving Performance in Reinforcement Learning by Breaking Generalization in Neural Networks

Toll-Based Learning for Minimising Congestion under Heterogeneus Preferences

In multiagent problems that require complex joint actions, reward shaping methods yield good behavior by incentivizing the agents' potentially valuable actions. However, reward shaping often requires access to the functional form of the reward function and the global state of the system. In this work, we introduce the Exploratory Gaussian Reward (EGR), a new reward model that creates optimistic stepping stone rewards linking the agents potentially good actions to the desired joint action. EGR models the system reward as a Gaussian Process to leverage the inherent uncertainty in reward estimates that push agents to explore unobserved state space. In the tightly coupled rover coordination problem, we show that EGR significantly outperforms a neural network approximation baseline and is comparable to the system with access to the functional form of the global reward. Finally, we demonstrate how EGR improves performance over other reward shaping methods by forcing agents to explore and escape local optima.

Downloads

Next from AAMAS 2020

Manipulation of Opinion Polls to Influence Iterative Elections

Similar lecture

Methods and Algorithms for Knowledge Reuse in Multiagent Reinforcement Learning

.css-70qvj9{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;}Downloads

Next from AAMAS 2020

Manipulation of Opinion Polls to Influence Iterative Elections

Similar lecture

Methods and Algorithms for Knowledge Reuse in Multiagent Reinforcement Learning

Downloads