EASSS 2024

Call For Tutorials

We invite tutorial proposals for the 23rd European Agent Systems Summer School, which will be hosted by School of Computer Science at University College Dublin, Ireland from August 19th until 23rd, 2024.

Since 1999, the annual European Agent Systems Summer School (EASSS) has provided a highly successful forum for knowledge exchange between researchers in the field of autonomous agents and multi-agent systems. As was the case with the earlier editions, EASSS 2024 will offer a rich program of both introductory and advanced courses on a broad range of topics in the area of Autonomous Agents and Multi-Agent Systems. The courses will be taught by leading researchers in the field, and are aimed at advanced Masters students, PhD students, and other young researchers.

We are interested in tutorial proposals in all areas of current research in Autonomous Agents and Multi-Agent Systems. As a very rough guideline, any topic that might be covered at the AAMAS conference or in the JAAMAS journal would be suitable for EASSS. We aim for a mix of tutorials on fundamental and well-established topics, as well as overviews of new and emerging areas of research. Tutorials should cover an appropriate selection of approaches and not specifically focus on the tutors' own contributions. Besides providing a coherent overview of a specific research topic, we specifically encourage tutorial proposals to articulate a clear link to applications and pragmatic consideration of the fundamental topics presented.

EASSS tutorials are typically 3 hours long, divided into two sessions of 90 minutes each. For the detailed call for tutorials, please see below. We encourage both well-established senior researchers and younger colleagues to submit proposals. Each tutorial is usually given by one or two tutors. Exceptions are of course possible; for proposals by more than two tutors, please justify this choice in the proposal and explain how you intend to ensure the coherence of the tutorial. To support the tutors, we can commit to the provision of up to 500 EUR per tutorial to cover the travel and accommodation costs.

Submission Instructions

A tutorial proposal should take the form of a single PDF document covering the points listed below. The maximum length of the proposal is 2 pages, and the use of our LaTeX template is mandatory.

The following points should be addressed in the tutorial proposal, preferably in the order stated below:

Title of the proposed tutorial.
Tutor(s): name, affiliation, full contact details, and a link to the personal homepage of each tutor.
Abstract: a short paragraph describing the tutorial (100-300 words), suitable for publication at the EASSS 2024 website.
Topics covered: detailed list of topics covered in the tutorial in terms of a list, preferably in the order in which the tutorial will cover them.
Type: a clear indication of whether the tutorial takes a theoretical or applications-oriented perspective on the topics covered. We encourage tutorial proposals that balance the two perspectives.
Level: please indicate the target audience and level of the course. If there are prerequisites, describe what knowledge will be assumed by the tutorial.
Teaching materials: Please indicate what kind of teaching materials you intend to provide. Note that we will require materials by the date above, so that they can be distributed to EASSS attendees in advance.
Duration: Tutorials are typically 3 hours long. If you have good reasons for wanting to deviate from this standard, please explain this in your proposal.
Equipment: Please list any special equipment (beyond data projector and whiteboard) that you might need.
Short biographical sketch for each tutor (around 100 words per tutor), suitable for publication at the EASSS 2024 website.
Experience of tutor(s): please provide details on relevant teaching experience of the tutor(s).
Additional information on the proposed tutorial. Include as much detail as you see fit.

Please submit the proposal (including LaTeX source) by email to rem.collier@ucd.ie mentioning [TUTORIAL] in the subject line.

Call For Participation

The 24th European Agent Systems Summer School (EASSS) will be held in the School of Computer Science of University College Dublin from August 19th until August 24th 2024.

This year, we are delighted to offer a rich programme of both introductory and advanced courses on a range of topics in the area of Autonomous Agents and Multi-Agent Systems including: Reinforcement Learning (RL); Coordination, Organisations, Institutions, Norms and Ethics (COINE), Engineering Multi-Agent Systems (EMAS) and Agent-Based Modelling and Simulation (ABMS). Further details can be found here.

In addition to these tutorials, we will invite topical keynote talks from leading researchers.

In addition to the tutorials, there will be an opportunity to to introduce yourself and your research work in a "Me-in-3" session. You will also be invited to engage in working groups to develop research ideas around a given theme. These ideas will be presented on the last day of EASSS and participants will be invited to co-author and publish (via arxiv.org) a paper based on their working group.

Finally, on Saturday 24th August, we will hold a school hackathon where teams will attempt to develop prototypes that will be presented. A prize will be awarded to the team that develops the best prototype.

Registration

We are still in the process of setting up the registration and payment system. Details of the registration fee can be found in the table below. Because EASSS 2024 is co-located with EUMAS 2024, we are delighted to offer a discounted joint registration fee to allow you to attend both events (EASSS+EUMAS). You can also attend the free workshop on Distributed Knowledge Graphs and Web Agents which will be held on the 29th August.

NEW! We are pleased to announce a number of student travel grants that will cover the costs of attending the summer school. To apply for a travel grant, please send a short statement of interest (why you want to attend the summer school) and confirmation that you have no other source of funding to rem.collier@ucd.ie. Applications will be assessed on a rolling basis until the funding runs out.

NOTE: You will need a letter from your host Institution clarifying that you have no alternative source of funding to access the grant.

	Until	Student (Fee)	Non-Student (Fee)
Early Bird:EASSS Only	10th August, 2024	€270	€300
Early Bird: EASSS + EUMAS	10th August, 2024	€450	€550
Normal: EASSS Only	19th August, 2024	€300	€350
Normal: EASSS + EUMAS	19th August, 2024	€500	€600

To register you intent to attend EASSS 2024, please go to the Payment Site. This site is a booking system hosted by University College Dublin that allows you to pay by credit or debit card. In order to make a reservation, you will need to create a UCD Connect Account. You can use this account to amend your booking if it is necessary.

If you require a letter for a visa, please complete the registration form and contact Prof. Rem Collier via email: rem.collier@ucd.ie.

Any additional queries can also be directed to this email address.

Programme

We are delighted to announce the programme for EASSS 2024. This years programme has been organised around 4 key research areas of the Multi-Agent Systems Community:

Applications of MAS
Reinforcement Learning (RL)
Coordination, Organisations, Institutions, Norms and Ethics (COINE)
Engineering Multi-Agent Systems (EMAS)
Agent-Based Modelling and Simulation (ABMS)

Each day will focus on one of these areas and will consist a keynote talk by a leading researcher together with one or two tutorials on that area.

Along side this, we will be running a series of working groups where school participants will work together to develop insights into a self selected research topic. At the end of the workshop, each group will present their findings. Following this, they will write a group report which will be published on arxiv.org and linked to from the school website.

Finally, we have also added an additional day to the workshop (Saturday 24th August) when we will run a hackathon challenge.

	Monday	Tuesday	Wednesday	Thursday	Friday	Saturday
9:00	Registration	COINE Keynote Prof. Maite Lopez Sanchez	RL Keynote Prof. Stefano V. Albrecht	EMAS Keynote Prof. Viviana Mascardi	ABMS Keynote Gary Polhill	Hackathon
9:30	Registration	COINE Keynote Prof. Maite Lopez Sanchez	RL Keynote Prof. Stefano V. Albrecht	EMAS Keynote Prof. Viviana Mascardi	ABMS Keynote Gary Polhill
10:00	Welcome	Tutorial 1: Deep RL	Tutorial 3: Current Trends in Argumentation Dynamics	Tutorial 4: Engineering Safe MAS	Tutorial 6: HPC & ABM
10:30	School Keynote: Prof. Munindar Singh
11:00	School Keynote: Prof. Munindar Singh
11:30	B R E A K
12:00	Me-in-3 Introduce yourself	Tutorial 1: Deep RL	Tutorial 3: Current Trends in Argumentation Dynamics	Tutorial 4: Engineering Safe MAS	Tutorial 6: HPC & ABM
12:30
13:00
13:30	L U N C H
14:00	L U N C H
14:30	Applications Keynote: Dr Panagiotis Tsarchopoulos	Tutorial 2: Continual Learning	Group Work	Tutorial 5: Game Theoretic Verification	Group Work
15:00	Applications Keynote: Dr Panagiotis Tsarchopoulos
15:30	Idea Generation
16:00	B R E A K		W R A P U P	B R E A K
16:30	Group Work	Tutorial 2: Continual Learning	Walking Tour of Dublin City	Tutorial 5: Game Theoretic Verification	Group Work & Presentations	Demonstrations & Prize Giving
17:00
17:30
18:00	W R A P U P			W R A P U P
Social:	Welcome Drinks	Irish Dancing	Irish Pub Visit	Dinner (T.B.C.)

Keynote Speakers

School Keynote: Sociotechnical Systems: History, Principles, and Programming
Professor Munindar Singh
A sociotechnical system (STS) comprises a social tier of stakeholders (people and organizations) and a technical tier of computational entities and resources. An STS exists to advance the requirements and values of its stakeholders and is governed through norms between them. STSs can be engineered or emergent and are usually both. The STS conception accords with many of our intuitions about the real world and it is obvious that multiagent systems would be a way to realize STSs. Yet it is not clear how to construct an STS. In the first part of the talk, I give a quick historical overview of multiagent systems (and what was called distributed AI) leading up to the modern era. I also describe how STSs relate to challenges and opportunities for research on autonomous agents and multiagent systems. In the second part of the talk, I describe research on STSs terms that avoids the limitations of current approaches, especially with respect to autonomy. I introduce interaction protocols and how to marry flexibility and rigor and describe how we can reason about norms and agent operations jointly.
Applications Keynote: Multi-Agent Systems in Smart Cities
Dr Panagiotis Tsarchopoulos
As urban environments evolve into highly interconnected and complex systems, Multi-Agent Systems (MAS) provide a robust framework for managing various smart city components. This tutorial provides an overview of how MAS can be applied across a wide array of smart city domains, including urban management, transportation, energy, public safety, smart homes and environmental monitoring. The session will delve into the practical aspects of integrating MAS into smart city infrastructure, supported by real-world examples and case studies that demonstrate implementations in various cities. Participants will gain an understanding of how MAS can optimize the functioning of these smart city systems, leading to more efficient, scalable, and resilient urban environments.
COINE Keynote: Inducing Ethical Behaviour in Autonomous Agents through Value Alignment
Professor Maite Lopez Sanchez
In this keynote, we explore how ethical values can shape the behaviour of autonomous agents. We begin with a concise introduction to the mathematical formalization of moral/ethical values, providing the foundation for their practical application in multi-agent systems. Next, we delve into various approaches to value alignment in the context of decision-making and learning processes. Special attention will be given to optimization techniques and reinforcement learning as powerful tools for aligning agent behaviour with ethical standards. Finally, we will demonstrate the application of these concepts through a series of simplified yet illustrative use cases, highlighting the potential of integrating ethics into autonomous decision-making.
RL Keynote: From Deep Reinforcement Learning to LLM-based Agents: Perspectives on Current Research
Associate Professor Stefano V. Albrecht
Since the recent successes of large language models (LLMs), we are beginning to see a shift of attention from deep reinforcement learning to LLM-based agents. While deep RL policies are typically learned from scratch to maximise some defined return objective, LLM-agents use an existing LLM at their core and focus on clever prompt engineering and downstream specialisation of the LLM via supervised and reinforcement learning techniques. In this talk, I will first provide a broad overview of my group's research in deep RL, which focuses among other topics on developing sample-efficient and robust RL algorithms for both single- and multi-agent control tasks, including industry applications in autonomous driving and multi-robot warehouses. I will then present our recent research into LLM-agents, where we propose an approach for household robotics that takes into account user preferences to achieve more robust and effective planning. I will conclude with some personal observations about the state of LLM-agent research: (a) many papers in this field follow essentially the same recipe by focussing on prompt engineering and downstream specialisation; (b) this recipe makes their scientific claims brittle as they depend crucially on the specific LMM engine, and (c) LLMs are not natively designed to maximise objectives for optimal control and decision making. Based on these observations, I believe some fruitful research avenues can be identified.
EMAS Keynote: Engineering Multiagent Systems: an introduction
Professor Viviana Mascardi
As any piece of software, agents and MAS must be modeled, designed, implemented, tested. In this talk, we will discuss the stages required for engineering MAS, along with tools and methods supporting them.
ABMS Keynote: Why is agent-based modelling such an awkward customer for institutional HPC?
Gary Polhill
This talk will briefly introduce agent-based modelling, which is a kind of computer simulation that explicitly represents the emergent dynamics from the interactions of heterogeneous interacting agents. Agents can represent people, but also proteins, plants, non-human animals, as well as households, businesses, and even nation states. After giving some examples of agent-based models from my own work, I will then show various ways in which high-performance computing (HPC) is, if not essential for studying agent-based models, then at least highly advantageous. However, the ways in which HPC access is institutionalized mean agent-based modellers are not typically using these resources, with work still stuck on personal computing devices. New solutions are needed, some of which might even entail some fun computing science.

Tutorials

Tutorial 1: Deep Reinforcement Learning: Foundations and Practical Environment Setup for Real- World Applications
This tutorial delves into the field of reinforcement learning (RL), a specialized area of machine learning focused on developing autonomous agents that behave optimally within specific environments. The tutorial will begin by covering fundamental RL concepts such as agents, environments, and learning paradigms, before discussing why deep learning is increasingly employed to solve complex, high-dimensional decision-making challenges. The session will introduce two principal types of deep RL algorithms: value-based and policy-based methods, each underscored by their respective applications and benefits. Following the theoretical overview, the tutorial will transition into a hands-on component where participants will learn to set up a training environment tailored for RL agents. This part of the tutorial will include a practical introduction to OpenAI’s Gym library and references to prominent libraries that implement deep RL algorithms. We will wrap up by tackling common deep RL challenges, such as managing the balance between exploration and exploitation and overcoming convergence issues. This comprehensive approach will provide attendees with the foundational knowledge necessary to initiate their own RL environments for custom projects.
Tutor: Franco Terranova, University of Lorraine
Tutorial 2: Continual Learning
Continual learning is a fundamental challenge in artificial intelligence, focusing on the ability of agents to learn continuously from a stream of data while retaining previously ac- quired knowledge. This tutorial provides a comprehensive overview of continual learning techniques, ranging from theoretical foundations to practical applications. Participants will gain insights into the latest advancements in this field and learn how to address the issues of catastrophic forgetting and model stability. Through hands-on exercises and case studies, attendees will explore how continual learning can be applied to real-world problems in various domains.
Tutor: Andrii Krutsylo, Polish Academy of Sciences
Tutorial 3: Current Trends in Argumentation Dynamics
Abstract argumentation is a simple model for representing conflicting pieces of information and reasoning from it. In particular, it can be used to model debates (e.g. for deliberation or negotiation) between several agents. In this tutorial, we will review the basic notions of abstract argumentation, and then focus on argumentation dynamics. This topic has received much attention in the last 15 years, with two main types of research questions: 1. given an information about the acceptability of arguments, how can we modify the argumentation framework to comply with this information? 2. given a mod- ification of the argumentation framework, what is the impact of this modification on the acceptability of arguments? The first question is related to strategic aspects (how can an agent convince her opponent to agree with her?), while the second question is mainly about computational aspects (how can we re-compute efficiently the acceptability status of arguments?). We will present the main contributions about both these questions, motivated by applications in the domain of (argument-based) automated negotiation.
Tutor: Jean-Guy Mailly, IRIT, University of Toulouse Capitole
Tutorial 4: Engineering Safe Multiagent Systems: an Introduction to Formal and Runtime Verifica- tion of Agents and MAS
In this tutorial we will discuss the importance of properly engineering agents and MAS, and we will introduce formal verification as a means to engineer safe systems. We will hence in- troduce the notions of formal and semi-formal verification, with special attention to runtime verification, and we will provide an overview on the literature on this topic, to clarify the context and the achieved results. Finally, we will go into the details of runtime verification of MAS, exploiting the RML Runtime Monitoring Language1 as a tool for showing demos and examples.
Tutor: Professor Viviana Mascardi, Professor Davide Ancona, Professor Angelo Ferrando
Tutorial 5: Game-Theoretic Verification of Multi-Agent Systems
Formal verification addresses the issue of correctness, which is a fundamental concern in computer science. In the context of multi-agent systems, rational verification is a game- theoretic counterpart of the conventional formal verification of computing systems. The rational verification problem is concerned with checking which properties will hold in a system when its constituent agents are assumed to behave rationally in pursuit of in- dividual objectives. In this tutorial, we will introduce the fundamental framework for rational verification, including concepts such as models of games, temporal logic, model and equilibrium checking problems, and the Reactive Modules modelling language. We will also present EVE, a formal verification tool that can be used to solve the most impor- tant decision problems in rational verification. We will show how we can model examples of multi-agent systems and use EVE to verify some given properties.
Tutor: Professor Muhammad Najib
Tutorial 6: High-Performance Computing with Agent-Based Models
This course will briefly introduce the steps you need to take to use your agent-based model on high-performance computing facilities. It focuses on agent-based models developed using NetLogo, and potentially other environments based on the Java Virtual Machine, but contains material relevant to any agent-based model in whatever language. It comprises a cut-down version of an earlier two-day tutorial on the same subject
Tutor: Gary Polhill

Local Information

Please find information on the venue and accomodation.

The Venue

UCD is one of Europe's leading research-intensive universities; an environment where undergraduate education, masters and PhD training, research, innovation and community engagement form a dynamic spectrum of activity. UCDs main campus is an extensive parkland estate of 133 hectares in South Dublin. Details of available transport links can be found here.

The venue for the summer school is the UCD School of Computer Science. This is labelled as building 18 on the map below. on the map below.

Leap Card

Leap Cards are a prepaid travel card that is the easiest way to pay your fare on public transport around Ireland. It is valid on most TFI services and commercial bus operators throughout Ireland. It’s more convenient because you don’t have to carry cash or queue at ticket machines and it can save you money because fares are usually up to 30% less than cash single tickets.

TFI 90 Minute Fare
When you have a leap card, you can travel anywhere in Dublin for 90 minutes for just €2. This includes all Dublin Bus, Go-Ahead Ireland, the Luas (tram) and the DART/Commuter Services (local train).

Accommodation Options

Much of the local accommodation is available through Booking.com. Occassionally, you can find better deals on their own websites.

On Campus:

	UCD Accommodation UCD has extensive student residences, but it is mostly rented out to groups attending English Language Camps over the summer. What remains is available through Booking.com at a rate of around €145 per day.

Nearby Hotels:

	Radisson St Helens (5 star) Closest external hotel to UCD. Very popular with limited availability. Prices start around €200 per night, but can rise quicky and steeply.

	Talbot Hotel, Stillorgan (4 star) This hotel is one of the closest hotels to UCD. It is around a 20-30 minute walk or around 10 minutes on a bus. Prices range from around €145 per night upwards depending on availability.

Alternatives:

	Premier Inn Hotel chain with accommodation at various locations. Rooms are €150+ per night. Would require a bus journey to get to UCD - duration depends on location.

	AirBnb AirBnb is very popular in Ireland and there are many options. Rooms can be rented for €80-100 per night. For those of you in groups, entire apartments can be rented for similar or better rates. Set "University College Dublin" as the location when you search.

	Generator Hostel A popular hostel in the city centre. Shared dormatories are available for around €50 per night. Bus to UCD is around 10-20 minutes walk. The bus takes 30-60 minutes depending on traffic.

	Jacobs Inn Hostel Cheaper alternative in the city centre. Around 30-60mins journey to UCD on the Dart or via Bus. Shared dormatories available from around €40 per night.

EASSS 2024

The 24th European Agent Systems Summer School

August 19-23rd, 2024 • Dublin, Ireland

About

Important Dates