Preliminary Schedule PDF ↓

08:30 - 09:00
Opening Remarks
09:00 - 13:00
13:00 - 15:00
Lunch Break
15:00 - 18:00
09:00 - 12:00
12:00 - 14:00
Lunch Break
14:00 - 17:00
17:00 - 18:30
Poster Session
09:00 - 12:00
12:00 - 14:00
Lunch Break
14:00 - 17:00
09:00 - 12:00
12:00 - 14:00
Lunch Break
14:00 - 15:30
15:30 - 16:00
Closing Remarks

Accepted Tutorials

Modern Recommender Systems 3 hours

Rodrigo da Silva Alves, Czech Technical University in Prague

This tutorial is designed to provide a comprehensive overview of recommender systems. It covers the basics of recommender system techniques, explores modern models for recommendation, and concludes with a discussion on the application of recommender systems under a multi-agent perspective.

Theory and Practice of Formal Argumentation 3 hours

Antonis Bikakis, University College London

Formal Argumentation is a research area within Artificial Intelligence, which uses formal methods from logic and graph theory to study arguments, argument-based inference and argument-based dialogues. Since Dung’s seminal paper “On the Acceptability of Arguments and Its Fundamental Role in Nonmonotonic Reasoning, Logic Programming, and n-Person Games” in 1995, which introduced Abstract Argumentation Frameworks (AAF), this area has been attracting growing interest, due to the simplicity of the AAF model, and its ability to capture various aspects of human reasoning and communication.

This tutorial will introduce the main concepts of formal argumentation and will provide an overview of the main developments in this area in the last three decades. Students will develop a good understanding of Abstract Argumentation Frameworks and extensions of this model that capture fundamental concepts of Multiagent Systems, such as dialogues, preferences and trust. Students will also develop a practical appreciation of Formal Argumentation with a hands-on session focused on its computational aspects and with a presentation of real-world applications in several domains such as the Web, Medicine and Law.

On the role of computational logic in MAS: practice with 2P-Kt 4 hours

Giovanni Ciatto, University of Bologna

Computational logic (CL) is a key enabler of intelligent behaviour for software agents. Theoretically, it supports key cognitive capabilities such as intensional knowledge representation, automated reasoning, deliberation, etc. When it comes to practice, logic-based technologies are often built on top, or as extensions of, the Prolog language—and the definite clauses (a.k.a. Horn clauses) logic (DCL) behind it. Stemming from a summary of the fundamental notions behind DCL, and logic unification and resolution, this tutorial discusses the impact of CL on the engineering of intelligent agents. Practical demonstrations and examples are provided by means of the 2P-Kt technology, which is a logic-based ecosystem where individual CL mechanisms can be exploited, combined, and possibly extended or modified to support agents’ cognitive capabilities—including, not limited to, automated reasoning.

Computation of Nash Equilibria: Recent Advances in Algorithms and Complexity 6 hours

Argyrios Deligkas, Royal Holloway University of London
Themistoklis Melissourgos, University of Essex

Nash equilirbium is the standard solution concept in game theory since it provides predictions for the behavior of interacting agents. One of the most fundamental problems in the field is to find a Nash equilibrium of a game. In this tutorial, we will cover the basics of game theory, through the lens of computer science, and we will revisit the recent advances on the field. Participants will learn the fundamental theory of two-player games and how to compute Nash equilibria in these games, will see several different kinds of many player games, and the computational challenges behind the computation of Nash equilibria.

Multi-Agent Oriented Programming 6 hours

Jomi Fred Hubner, Federal University of Santa Catarina
Luis Gustavo Nardin, École Nationale Supérieure des Mines de Saint-Étienne
Andrei Ciortea, University of St.Gallen

This tutorial provides an integrated view of the four complementary dimensions of an approach to Multi-Agent Oriented Programming (MAOP): agent, interaction, environment, and organization. We illustrate the approach by using the JaCaMo framework, which provides the means to seamlessly integrate all these dimensions. JaCaMo has already been developed for a certain number of years and is fairly robust and fully-fledged. The tutorial introduces the relevant MAOP’s concepts and techniques through an illustrative scenario that simultaneously uses all four dimensions. Hands-on activities are proposed for each dimension to endow participants with the practical skills on programming MAS.

Preference-based argumentation in practice 6 hours

Nikolaos Spanoudakis, Technical University of Crete

The tutorial will motivate the use of argumentation and defeasible logic. Then it will outline Dung’s abstract argumentation, followed by a structured argumentation framework (Gorgias). The presentation will relate structured argumentation to explainable AI and the attributive, contrastive and actionable properties of an explanation. It will close with a short presentation of two real-life use cases where cognitive argumentation technology is used at the center of these applications to provide an explainable and Human-centric solution. The software methodology to acquire the knowledge for these applications and the use of the Gorgias system tools will be briefly presented. In a second part, the students can have the possibility to do a hands on session to the Gorgias Cloud system to develop and test their own theories in a specific decision making problem.

Multi-Agent Path Finding 90 minutes

Pavel Surynek, Czech Technical University in Prague

Multi-Agent Path Finding (MAPF) is the problem of computing collision-free paths for a team of agents from their current locations to given goal locations Application examples include automated warehouse systems, autonomous cars or UAVs, and game characters in video games. Practical systems must find high-quality collision-free paths for such agents quickly. This often means to minimize the objective such as the sum-of-costs, that is the total number of actions performed by the agents, or the makespan, that is, the number of time steps until all agents are in their goal locations. In this tutorial, we will give overview of methods for MAPF based on heuristic search and reduction of the problem to Boolean satisfiability.

Autonomous agents and agent based systems applied to Bond Markets: Can we build a better market using AI? 3 hours

Alicia Vidler, University of New South Wales

Agent based modeling used for financial markets is still very much in its infancy. Historically, financial market modeling has concentrated around stochastic calculus based models used to predict asset prices (such as Black Scholes etc) or game theory applied to auction processes in transparent centralised public asset exchanges (such as the New York stock exchange). However, a significant portion of the worlds’ retirement-funds are placed in government bonds issues by countries such as American, UK, Australia and Japan. These bonds do not trade on the New York stock exchange, or typically, on any such public transparent market. They trade through a variety of human and electronic networks, often with very littler formal centralisation or visability. Rules governing these marketplaces are implemented through central bank regulation, not stock exchanges. The bond market I refer to is estimated to be around 3 times the size of all the well known equity markets combined. Approximately 50% of pension retirement savings globally are investing in such bonds, which need to trade in order to be turned into cash to pay retirement liabilities. The finance industry is eager to see "AI" applied, but little academic work has made in roads into how this might work for assets other than equity markets. I will explore the techniques, challenges and requirements needed to apply agent based modeling methods to such systematically important bond markets. My tutorial will also necessarily discuss features and aspects of what makes bond markets both unique, and why developing AI and, in particular, explainable AI methods, for markets have application to other areas of finance.