INFORMED AI Launch Day – Monday 21 October 2024
We are pleased to announce the Launch of the INFORMED-Hub at an event in the MShed in Bristol on 21 October.
INFORMED-AI looks to establish theory and algorithms that underpin state-of-the-art AI in networks of heterogeneous agents, enabling the design of ecosystems in which agents – both natural and artificial – collaborate, compete, influence, and learn from each other.
Our academic and industry researchers will showcase our research themes and the work they hope to undertake during this project.
For those interested in the mathematical foundations of AI or Machine Learning (academic or industry based), we invite you to join us at this event.
Draft schedule:
10:00 AM Arrival and coffee
10:15 AM Intro and Welcome Sidharth Jaggi
10:30 AM Scheduling In Quantum Networks Neil Walton/ Thiru Vasantam
Abstract: We assess the performance of quantum switching technologies. Such analysis is becoming increasingly important as the number of quantum networks in operation increases. Quantum switches play a crucial role in these networks by generating, distributing, and managing entanglements. Unlike traditional switching systems, quantum switches operate as a two-sided queuing network and produce Link-Level Entanglements (LLEs). These LLEs are then combined to fulfil the network’s requests for entanglement. We have been able to identify an optimal the scheduling algorithm, which on a fast-timescale must solve a specific average reward Markov Decision Process. This insight opens the possibility to distributed optimization quantum switching technologies with reinforcement learning techniques.
11:00 AM Maximal correlation and the information-theoretic Central Limit Theorem Oliver Johnson
Abstract: I will explain how ideas from information theory can help shed light on statistical problems. I will focus on the Central Limit Theorem, showing how Barron’s original proof of convergence in KL divergence has been extended to give explicit rates of convergence. By understanding this in terms of projections and eigenfunction problems, I will show how this relates to the concept of maximal correlation, a way of measuring dependence which can be useful in many AI settings.
11:30 AM 15 minute break
11:45 AM Heterogeneous Thresholds, Social Ranking and the Emergence of Vague Categories Jonathan Lawry
Abstract: Threshold models in which an individual’s response to a particular state of the world depends on whether or not an associated measured value exceeds a given threshold, are common in a variety of social learning and collective decision making scenarios in both natural and artificial systems. If thresholds are heterogeneous across a population of agents then graded population level responses can emerge in a context in which individual responses are discrete and limited. In this paper we propose a threshold-based model for social learning of shared quality categories. This is then combined with the voting model of fuzzy categories to allow individuals to learn membership functions from their peers which can then be used for decision making including ranking a set of available options. We use agent-based simulation experiments to investigate variants of this model and compare them to an individual learning benchmark when applied to the ranking problem. These results show that a threshold-based approach combined with category based voting across a social network provides an effective social mechanism for ranking which exploits emergent vagueness.
12:15 PM Summary and discussion
12:30 PM Lunch and networking
13:30 PM Industry research talks:
Chris Swinerd (Dstl Fellow, Visiting Professor of Electromagnetic Systems, Wolfson School, Loughborough University) INFORMED-AI CDT Launch Event – Introduction to Dstl
Abstract: This short presentation introduces Dstl, its strategic S&T capabilities and outlines the focus of its AI Programme. Without commitment, the presentation aims to signpost the potential scope for Dstl support to the INFORMED-AI CDT.
Joseph Tedds (Senior Quantum Algorithm Engineer, Cambridge Consultants) Heterogenous approaches to QML: Less quantum, more information
14:30 PM 30 min break
15:00 PM Keynote
Laurent Massoulie (Scientific Director, Paris Inria Centre) Graph Alignment: Informational and Computational limits
Abstract: Graph alignment is a generic unsupervised learning task with many applications, from neuroscience to social network de-anonymization. The alignment of a pair of correlated random graphs turns out to be a problem in high dimensional statistics featuring a rich set of phenomena. Specifically, we shall first present results on information-theoretic limits to feasibility of graph alignment: below some threshold, the observed graphs do not contain enough information for alignment to be feasible, while above that threshold, some algorithms are known to succeed at alignment, although in exponential time. We shall then present results on computational feasibility of alignment, describing a second threshold which determines when a family of ‘local’, polynomial-time algorithms succeed at alignment. Together, these results show a rich phenomenology for graph alignment, displaying an ‘impossible phase’, a ‘hard phase’ and an ‘easy phase’ for feasibility of the task.
16:00 PM Networking opportunity
17:00 PM Close
Code of Conduct in line with University of Bristol School of Mathematics
Keynote Speaker:
Laurent Massoulié is research scientist at Inria, scientific director of the Paris Inria Centre and professor at the Applied Maths Centre of Ecole Polytechnique. His research interests are in machine learning, probabilistic modelling and algorithms for networks. He has held research scientist positions at: France Telecom, Microsoft Research, Thomson-Technicolor, where he headed the Paris Research Lab. He obtained best paper awards at IEEE INFOCOM 1999, ACM SIGMETRICS 2005, ACM CoNEXT 2007, NeurIPS 2018, NeurIPS 2021, was elected “Technicolor Fellow” in 2011, received the “Grand Prix Scientifique” of the Del Duca Foundation delivered by the French Academy of Science in 2017, is a Fellow of the “Prairie” Institute and received the ACM Sigmetrics achievement award in 2023.
Hotels near to the venue (if required)
Travelodge Bristol Central Hotel
The EPSRC CDT in Computational Statistics and Data Science is holding its annual conference in the same venue on Tuesday 22 October 2024. The focus this year is “The Future of Data Science” and includes a keynote by Prof Aline Villavicencio, Director of the Institute of Data Science and Artificial Intelligence, University of Exeter. This is an opportunity to hear from PhD students and academic staff about the research being undertaken in the Institute for Statistical Science at the University of Bristol. Please indicate via the email invitation and registration form you receive if you would like to attend the Compass CDT Conference on 22 October 2024.