New Zealand

What goals should a multi-goal reinforcement learning agent pursue during training in long-horizon tasks? When the desired (test time) goal distribution is too distant to offer a useful learning signal, we argue that the agent should not pursue unobtainable goals. Instead, it should set its own intrinsic goals that maximize the entropy of the historical achieved goal distribution. We propose to optimize this objective by having the agent pursue past achieved goals in sparsely explored areas of the goal space, which focuses exploration on the frontier of the achievable goal set.We show that our strategy achieves an order of magnitude better sample efficiency than the prior state of the art on long-horizon multi-goal tasks including maze navigation and block stacking.

AAMAS 2020

Maximum Entropy Gain Exploration for Long Horizon Multi-goal Reinforcement Learning

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). 

The purpose of the AAAI conference series is to promote research in Artificial Intelligence (AI) and foster scientific exchange between researchers, practitioners, scientists, students, and engineers across the entirety of AI and its affiliated disciplines. AAAI-23 is the Thirty-Seventh AAAI Conference on Artificial Intelligence. The theme of this conference is to create collaborative bridges within and beyond AI. Like previous AAAI conferences, AAAI-23 will feature technical paper presentations, special tracks, invited speakers, workshops, tutorials, poster sessions, senior member presentations, competitions, and exhibit programs, and two new activities: a Bridge Program and a Lab Program. Many of these activities are tailored to the theme of bridges and all are selected according to the highest standards, with additional programs for students and young researchers. 
AAAI is providing you with a conference planner, which you can use to help organize your itinerary of activities. This includes talks to attend in person, talks to attend remotely, breaks with colleagues and your site seeing activities. To access this conference planner, please go to [https://aaai-2023.takemobi.io](https://aaai-2023.takemobi.io).

In order to access this site, you need to register. If you haven't already, please register [here](https://aaai.org/Conferences/AAAI-23/registration/).


AAAI 2023

The purpose of the AAAI conference series is to promote research in Artificial Intelligence (AI) and foster scientific exchange between researchers, practitioners, scientists, students, and engineers across the entirety of AI and its affiliated disciplines.

technical paper

Encapsulating Reacting Behaviour In Goal-Based Plans For Programming BDI Agents

Interactive Narrative (IN) is an emerging application of planning for control of virtual character behaviours. Despite popularity in research, more widespread adoption of planning has been hindered by the difficulty of authoring planning domain models as INs require models which are robust to dynamic environments and capable of generating a diversity of solutions. To reduce this authoring burden we explored automated extension of partially developed planning domains to address robustness and diversity. We introduce a novel algorithm, THYPE, which proposes additional types of characters and objects to extend a domain model. The extensions increase robustness by enlarging the range of ways to enact planned behaviours. In the paper we embed THYPE within a modular and extensible framework that allows multiple off-line generated extensions to be combined. We present results of a user study that show THYPE suggestions are plausible. We also empirically demonstrate the ability of THYPE extensions to increase the robustness of the models, demonstrate the modular nature of our extension framework by combining THYPE with another extension approach dedicated to recovery from plan failure and present results which show that enhanced performance results from combining multiple extensions


Extending Narrative Planning Domains with Linguistic Resources

In-time Explainability in Multi-Agent Systems: Challenges, Opportunities, and Roadmap

Welcome to EMAS 2020

SARL is a general-purpose agent-oriented programming language. This language aims at providing the fundamental abstractions for dealing with concurrency, distribution, interaction, decentralisation, reactivity, autonomy and dynamic reconfiguration that are usually considered essential for implementing agent-based applications. Every programming language specifies an execution model. In the case of SARL, this execution model is defined based on the object-oriented paradigm, i.e., the run-time environment is written in Java. Accordingly, and by default, the SARL programs are transformed into their equivalent object-oriented programs written in Java. The goal of this talk is twofold. Firstly, to explain the mapping between the agent-oriented paradigm and the object-oriented paradigm. Secondly, to define transformations from the SARL constructs to the Java constructs. We advocate that these goals enable the SARL developers to understand the SARL statements, and the mapping to the equivalent executable entities.

Model Transformations from the SARL Agent-Oriented Programming Language to an Object-Oriented Programming Language

iLQGames.jl: Rapidly Designing and Solving Differential Games in Julia

Learning Cooperative Solution Concepts from Voting Behavior

ORTHOS: A Trustworthy AI Framework for Data Acquisition

Modern interactive software, such as computer games, employ complex user interfaces. Although these user interfaces make the games attractive and powerful, unfortunately they also make them extremely difficult to test. Not only do we have to deal with their functional complexity, but also the fine grained interactivity of their user interface blows up their interaction space. A space so complex, that traditional automated testing techniques have trouble handling it. An agent-based testing approach offers an alternative solution. The agents' goal driven planning, adaptivity, and reasoning ability can provide an extra edge towards effective navigation in complex interaction space. This paper presents aplib, a Java library for programming intelligent test agents, featuring novel tactical programming as an abstract way to exert control over agents' underlying reasoning-based behavior. This type of control is suitable for programming testing tasks. Aplib is implemented in such a way to provide the fluency of a Domain Specific Language (DSL). Its embedded DSL approach also means that aplib programmers will get all the advantages that Java programmers get: rich language features and a whole array of development tools.

Aplib: Tactical Agents for Testing Computer Games

Node similarity measures quantify how similar a pair of nodes are in a network. These similarity measures turn out to be an important fundamental tool for many real world applications such as link prediction in networks, recommender systems etc. An important class of similarity measures are local similarity measures. Two nodes are considered similar under local similarity measures if they have large overlap between their neighboring set of nodes. Manipulating node similarity measures via removing edges is an important problem. This type of manipulation, for example, hinders effectiveness of link prediction in terrorists networks. All the popular computational problems formulated around manipulating similarity measures turn out to be NP-hard. We, in this paper, provide fine grained complexity results of these problems through the lens of parameterized complexity. In particular, we show that some of these problems are fixed parameter tractable (FPT) with respect to various natural parameters whereas other problems remain intractable (W[1]-hard and W[2]-hard in particular). Finally we show the effectiveness of our proposed FPT algorithms on real world datasets as well as synthetic networks generated using Barabasi-Albert and ErdosRenyi models.

Maximum Entropy Gain Exploration for Long Horizon Multi-goal Reinforcement Learning

Downloads

Next from AAMAS 2020

Encapsulating Reacting Behaviour In Goal-Based Plans For Programming BDI Agents

Similar lecture

OPT-GAN: A Broad-Spectrum Global Optimizer for Black-box Problems by Learning Distribution

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES

Maximum Entropy Gain Exploration for Long Horizon Multi-goal 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

Encapsulating Reacting Behaviour In Goal-Based Plans For Programming BDI Agents

Similar lecture

OPT-GAN: A Broad-Spectrum Global Optimizer for Black-box Problems by Learning Distribution

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES

Downloads