PRIMA 2025

Affiliation: Kyoto University, Japan

Talk Title: HyperDemocracy: Large-scale Consensus Support Platform based on Multiagent Systems

Date & Time: TBA

Abstract:

The rapid development of AI, smartphones, and the Internet, as well as the recent changes in the environment caused by COVID-19, have made the realization of AI-powered social systems more and more realistic. We envision the Hyperdemocary Platform, a social network that supports software agents and humans to participate, discuss, and make a collaborative, democratic consensus. Specifically, multiple agents will be stationed in social networks as the platform of democracy, act with humans, mediate decisions and interactions, and support enhanced and scalable consensus-building and collective decision-making. In particular, while collecting opinions and preferences, software agents are attentive to the emotions of the participants and help them to reach a consensus that is both proactive and satisfying efficiently. In other words, the system solves the trade-off between proactive and satisfied consensus building by humans and super-efficient and super-rational consensus building by software agents. Social problems such as flaming, fake news, group polarization, and gerrymandering have been widely reported in the SNS. Our agents will solve these problems by collaborating with humans within the platform. In this talk, I will introduce some results of our recent work.

Biography:

Takayuki Ito is Professor of Kyoto University. He received the B.E., M.E, and Doctor of Engineering from the Nagoya Institute of Technology in 1995, 1997, and 2000, respectively. From 1999 to 2001, he was a research fellow of the Japan Society for the Promotion of Science (JSPS). From 2000 to 2001, he was a visiting researcher at USC/ISI (University of Southern California/Information Sciences Institute). From April 2001 to March 2003, he was an associate professor of Japan Advanced Institute of Science and Technology (JAIST). From April 2004 to March 2013, he was an associate professor of Nagoya Institute of Technology. From April 2014 to September 2020, he was a professor of Nagoya Institute of Technology. From October 2020, he is a professor of Kyoto University. From 2005 to 2006, he is a visiting researcher at Division of Engineering and Applied Science, Harvard University and a visiting researcher at the Center for Coordination Science, MIT Sloan School of Management. From 2008 to 2010, he was a visiting researcher at the Center for Collective Intelligence, MIT Sloan School of Management. From 2017 to 2018, he was an invited researcher of Artificial Intelligence Center of AIST, JAPAN. From March 5, 2019, he is the CTO of AgreeBit, inc. He is Steering Committee Chair of PRIMA, Steering Committee Member of PRICAI, Executive Committee Member of IEEE Computer Society Technical Committee on Intelligent Informatics, the PC-chair of AAMAS2013, PRIMA2009, the Local Arrangements Chair of IJCAI-PRICAI2020, General-Chair of PRICAI2024, PRIMA2024, PRIMA2020, PRIMA2014, and was a SPC/PC member in many top-level conferences (IJCAI, AAMAS, ECAI, AAAI, etc). He was a board member of IFAAMAS. He received the JSAI (Japan Society for Artificial Intelligence) Contribution Award, the JSAI Achievement Award, the JSPS Prize, 2014, the Prize for Science and Technology (Research Category), The Commendation for Science and Technology by the Minister of Education, Culture, Sports, Science, and Technology, 2013, the Young Scientists' Prize, The Commendation for Science and Technology by the Minister of Education, Culture, Sports, Science, and Technology, 2007, the Nagao Special Research Award of the Information Processing Society of Japan, 2007, the Best Paper Award of AAMAS2006, the 2005 Best Paper Award from Japan Society for Software Science and Technology, the Best Paper Award in the 66th annual conference of 66th Information Processing Society of Japan, and the Super Creator Award of 2004 IPA Exploratory Software Creation Projects. He is Principle Investigator of the Japan Cabinet Funding Program for Next Generation World-Leading Researchers (NEXT Program). Further, he has several companies, which are handling web-based systems and enterprise distributed systems. His main research interests include multi-agent systems, intelligent agents, collective intelligence, group decision support system, etc.

Affiliation: Università degli Studi di Torino, Italy

Talk Title: Programming Interaction-Oriented Cognitive Agents

Date & Time: TBA

Abstract:

In 2012, Michael Winikoff highlighted two key shortcomings of agent-oriented programming languages (AOPLs). First, despite the importance of modeling interactions in multiagent systems (MAS), AOPLs provided little more than basic primitives for sending and receiving messages. He viewed this reliance on low-level primitives as a form of control transfer between agents and drew an unfavorable analogy to the use of goto statements in traditional programming. Second, interaction protocols—typically expressed in notations such as AUML—were message-centric and overly constrained agent interactions. To promote robustness and flexibility in interactions, as demanded by agent autonomy, Winikoff advocated for higher-level abstractions that conceal low-level messaging details. As of 2025, AOPLs continue to suffer from the shortcomings Winikoff identified. In response, we have developed Orpheus and Azorus, which enable a programming model for multiagent systems that integrates cognitive abstractions with high-level, flexible models of agent interaction. Our approach captures the semantics of interaction through the specification of commitments and models the operational constraints of interaction through information protocols. Built on top of Jason, Orpheus and Azorus fully leverage an agent’s cognitive autonomy—based on its goals, beliefs, and intentions. They support reasoning about both goals and commitments, and they integrate the modeling of commitments with interaction protocols. In doing so, Orpheus and Azorus unify three leading declarative approaches to engineering decentralized multiagent systems.

Biography:

Matteo Baldoni is a professor at the University of Turin, Department of Computer Science. He received a Ph.D. in Computer Science in 1998. He has co-authored more than 150 papers in the field of Artificial Intelligence, particularly in Computational Logic, the Semantic Web, Multi-Agent Systems, and Business Processes. He has served as co-chair of numerous events in the AI field, including EUMAS 2025 (European Conference on Multi-Agent Systems), PRIMA 2016 (Principles and Practice of Multi-Agent Systems), and AIxIA 2013 (Advances in Artificial Intelligence), as well as five editions of the International Workshop on Declarative Agent Languages and Technologies and three editions of the International Workshop on Engineering Multi-Agent Systems. He was General Chair of PRIMA 2019 and AIxIA 2020.

Affiliation: Imperial College London, UK

Talk Title: Towards verifying neural-symbolic multi-agent systems

Date & Time: TBA

Abstract:

A challenge in the deployment of multi-agent systems (MAS) remains the inherent difficulty in confidently predicting their run-time behaviour. Over the past twenty years, increasingly scalable verification methods, including model checking and parameterised verification, have enabled the validation of several classes of MAS against AI-based specifications, and several MAS applications in services, robotics, security, and beyond. Yet, a new class of agents is emerging in applications. Differently from traditional MAS, which are typically directly programmed (and less often purely neural), they combine both connectionist and symbolic aspects. We will refer to these as neural-symbolic MAS. These agents include a neural layer implementing their perception function and symbolic or control-based layers realising decision making and planning. Implementations of neural-symbolic agents permeate many present and forthcoming AI applications, including autonomous vehicles and robotics. Due to the neural components, as well as their heterogeneity, verifying the behaviours of neural-symbolic MAS is particularly challenging. Yet, I will argue that, given the safety-critical applications they are used in, it is important we develop methods and tools to address their formal verification. In this talk I will share some of the contributions on this topic developed in our Lab at Imperial College London. I will begin by describing traditional approaches for the verification of symbolic MAS, and parameterised verification to address arbitrary collections of agents such as swarms. I will then summarise our present efforts on verification of neural perception systems, including MILP-based approaches, linear relaxations, and symbolic interval propagation, introduce our resulting toolkits, Venus and Verinet, and exemplify their use. This will lead us to existing methods for closed-loop, neural-symbolic MAS. In this context, I will share existing results that enable us to perform reachability analysis, and verify systems against bounded temporal specifications and ATL. I will conclude by highlighting the challenges that in my view lie ahead.

Biography:

Alessio Lomuscio (http://www.doc.ic.ac.uk/~alessio) is Professor of Safe Artificial Intelligence in the Department of Computing at Imperial College London (UK), where he leads the Safe AI Lab (http://sail.doc.ic.ac.uk/). He is a Distinguished ACM member, a Fellow of the European Association of Artificial Intelligence and currently holds a Royal Academy of Engineering Chair in Emerging Technologies. He was a founding co-PI for the UKRI Doctoral Training Centre in Safe and Trusted Artificial Intelligence. Alessio's research interests concern the development of formal verification methods for artificial intelligence. Since 2000 he has worked on the development of formal methods for the verification of autonomous systems and multi-agent systems. To this end, together with colleagues he has put forward several methods based on model checking and various forms abstraction to verify AI systems, including robotic swarms, against AI-based specifications. He is presently focusing on the development of methods to ascertain the correctness of AI systems based on deep neural networks.