This workshop aims to enhance interdisciplinary dialogue on Collective Intelligence (CI), focusing on "Exploring new frontiers" in Artificial Life research across three areas: Evolution, Criticality, and Creativity. CI, the intelligent behaviour emerging from collective efforts, plays a crucial role in various domains. The event encourages participation from diverse fields—biology to sociology and art—to bring fresh insights into ALIFE conferences. It will delve into how evolution, criticality, and creativity intersect within collective systems to foster CI.
13:00 - 13:10 Opening remarks
13:10 - 13:55 Andreagiovanni Reina
Title: The power of inhibition for collective decision making in minimalistic robot swarms
Abstract:
I investigate how large groups of simple robots can reach a consensus with decentralised minimalistic algorithms. Simple robots can be useful in nanorobotics and in scenarios with low-cost requirements. I show that through decentralised voting algorithms, swarms of minimalistic robots can make best-of-n decisions. In my research, I show that using a biologically-inspired voting model based on inhibitory signals, the swarm can collectively perform better and be more resilient against a minority of misbehaving robots than in models without inhibition. Animal behaviour is the source of inspiration for robot algorithms; in turn, the resulting large-scale robotics experiments and mathematical models can help better understand the natural systems and the evolutionary advantages of certain signals. Collective behaviour can be counterintuitive, and as Anderson said in 1972, More Is Different. I will also show some counterintuitive results where the stereotypical view that noise is something to suppress is broken; on the contrary, there are relevant conditions where noise and individual errors can be beneficial components to the group behaviour, improving group accuracy and responsiveness to environmental changes.
Speaker's Bio:
Andreagiovanni Reina is a Research Group Leader of the Centre for the Advanced Study of Collective Behaviour of the University of Konstanz and Max Planck Institute of Animal Behavior, Germany. His research is highly interdisciplinary in both its scope and its methodology, with numerous contributions to a variety of disciplines, including computer science, robotics, theoretical biology, physics, cognitive neuroscience and psychology. Andreagiovanni's interdisciplinary approach consists of combining techniques from dynamical systems theory, statistical physics, network science, statistical optimality theory, multiagent simulation and large-scale robotics. From 2021 to 2023, Andreagiovanni was a Research Fellow in Collective Behaviour at the Interdisciplinary Institute for Artificial Intelligence (IRIDIA) of the Université Libre de Bruxelles, funded by the Belgian F.R.S.-FNRS as a Chargé de Recherches. Previous to that, from 2015 to 2020, Andreagiovanni was a Research Fellow at the University of Sheffield (UK). He holds a PhD in Applied Sciences from IRIDIA, Université Libre de Bruxelles, and an MSc in Computer Engineering from Politecnico di Milano, Italy. He has been a researcher in several European projects on distributed robotic systems since 2009.
13:55 - 14:10 Alife for AI for Collective Intelligence | Seth Bullock (University of Bristol):
14:15 - 14:30 Foragax: A Multi-Agent Foraging Toolkit | Siddharth Chaturvedi, Ahmed El-Gazzar and Marcel van Gerven (Radboud University)
14:30 - 15:00 Coffee break
15:00 - 15:45 Joachim Winther Pedersen
Title: Neurons, Synapses, Meta-Reinforcement Learning and Collective Intelligence
Abstract: Biological neurons from a diverse set of classes connect and communicate with each other to some of the most complex systems in existence. Mimicking the self-organizing capabilities of biological brains in simulation remains a significant outstanding challenge for the field of computational neuroscience. Achieving networks that can quickly self-organize and adapt is also desirable within the field of artificial intelligence (AI) as a means to overcome the challenges associated with the rigidness of artificial neural networks (ANNs).
This talk will focus on research that takes the perspective that ANNs can be made more general and flexible by optimizing neural building blocks, independent from a fixed network structure. While this direction is often framed as falling within the field of meta-learning, the shifted focus onto the shared intelligence that emerges from the interactions of several individuals directly aligns with the field of collective intelligence. We will discuss some of the shared challenges faced by neural building blocks and multi-agent reinforcement learning, such as the symmetry dilemma and the credit assignment problem.
15:45 - 16:00 Evolving Creativity: A Novel Approach to Musical Time Series Forecasting through Tangled Program Graphs | Ali Naqvi and Stephen Kelly (McMaster University)
16:00 - 16:15 Resist/Influence! Evolving for Co-existing Subcultures in Robot Swarms | Khulud Alharthi, Suet Lee, Simon Jones and Sabine Hauert (University of Bristol)
16:15 - 16:30 Closing remarks
We invite researchers and practitioners to submit papers that explore the following (and related) areas:
Autonomous decentralized collective robotics and machine learning methods inspired by biology
Research in anthropology, evolutionary and social psychology through collective reinforcement learning and multi-agent simulation
Intervention in human groups by artificial robots or agents
Theoretical and experimental research on animal collective behavior
Using game theory in economic research.
Neuroscience focusing on the criticality of neural activity
Other approaches to expand the horizon of collective intelligence research
Original research, summary of published results and opinions are welcome.
Submissions should be in the form of extended abstracts (maximum 2 pages) or research articles (maximum 8 pages). If your contribution is a summary of previous results, please specify during the submission process.
Accepted work will be made available in the format of a booklet.
Deadline for Abstract Submission: 3rd June
Notification of Abstract Submission: 10th June
Presentation Time: Each selected speaker will be allocated 15 minutes (12 mins talk + 3 mins Q&A) for their presentation.
Please send the title and abstract of your presentation via email to the event organisers at ECCCI.ALIFE@gmail.com.
Social/Evolutionary Psychology
Machine learning and robotics in collective intelligence
Criticality in distributed system
Creativity in group dynamics
The University of Konstanz
IT University of Copenhagen
Kazuya Horibe (RIKEN CBS | Osaka University)
Michael Crosscombe (The University of Tokyo)