On Wednesday 16 April, we have an enhanced version of our AI Fun with ELLIS seminar, with two visiting speakers - Prof Andy Cooper and Dr Xenofon Evangelopoulos - from the University of Liverpool. You can find the details of the talks they will deliver below. This event will include a catered break in between the two talks.
Speaker: Prof Andrew Cooper, Materials Innovation Factory and Department of Chemistry, University of Liverpool
Talk Title: Putting a Brain in the Mobile Robotic Chemist
Abstract: I will discuss our development of the mobile robotic chemist—an autonomous robot that can carry of chemical experiments, much like a human chemist, making its own decisions. The lecture will chart the evolution of this work from catalysis research 1 to crystalline materials 2 and exploratory organic synthesis 3. I will also discuss our more recent work where we have sought to incorporate human knowledge 4 and, most recently, reasoning from large language models 5 into such autonomous workflows. Our long-term goal is to create robots that can go beyond multivariate optimization problems and invent new chemistry and materials using hybrid human-AI reasoning.
1 A mobile robotic chemist, B. Burger, P. M. Maffettone, V. V. Gusev, C. M. Aitchison, Y. Bai, X. Wang, X. Li, B. M. Alston, B. Li, R. Clowes, N. Rankin, B. Harris, R. S. Sprick & A. I. Cooper, Nature 583, 237–241 (2020)
2 Modular, multi-robot integration of laboratories: an autonomous workflow for solid-state chemistry, A. M. Lunt, H. Fakhruldeen, G. Pizzuto, L. Longley, A. White, N. Rankin, R. Clowes, B. Alston, L. Gigli, G. M. Day, A. I. Cooper & S. Y. Chong, Chem. Sci. 15, 2456–2463 (2024)
3 Autonomous mobile robots for exploratory synthetic chemistry, T. Dai, S. Vijayakrishnan, F. T. Szczypi?ski, J.-F. Ayme, E. Simaei, T. Fellowes, R. Clowes, L. Kotopanov, C. E. Shields, Z. Zhou, J. W. Ward & A. I. Cooper, Nature 635, 890–897 (2024)
4 HypBO: Accelerating black-box scientific experiments using experts’ hypotheses,
A. Cissé, X. Evangelopoulos, S. Carruthers, V. V. Gusev & A. I. Cooper, IJCAI, 3881–3889 (2024)
5 Language-based Bayesian optimization research assistant (BORA), A. Cissé, X. Evangelopoulos, V. V. Gusev & A. I. Cooper https://arxiv.org/abs/2501.16224
Acknowledgements: AIC thanks the Royal Society for a Research Professorship. We thank the Leverhulme Trust for supporting the Leverhulme Research Centre for Functional Materials Design. We thank EPSRC for funding AI for Chemistry (AIchemy, EP/Y028775/1 and EP/Y028759/1).
Tea/Coffee Break
Speaker: Dr Xenofon Evangelopoulos, Leverhulme Research Centre for Functional Materials Design, University of Liverpool
Talk Title: Human-informed AI for ‘in-the-loop’ Chemical Discovery
Abstract: Global challenges such as sustainability and antibiotic resistance stress the demand for rapid development of new materials and drugs. For chemists, the design pipeline is prohibitively slow involving extensive literature reviewing, laborious experimentation and complex decision-making. Sample-based optimisation methods can help 'model' the design space and simultaneously estimate the uncertainty of multiple design outcomes fast, however, chemical design spaces are yet dauntingly high-dimensional, non-convex and with often isolated fruitful regions, calling for the need of more 'chemically contextualised' searches. In this talk I will discuss how chemical knowledge can be injected in a Bayesian optimisation loop, in various forms, i.e., literature, human feedback, simulations etc. to help further accelerate and improve the search. I will also discuss how state-of-the-art ML such as Large Language, and Multi-modal Transformers can revolutionise how chemical knowledge can be captured and injected in the loop and conclude my talk with open challenges in that direction.
If you are unable to join us in person, you may join us online via Zoom: if you wish to do this, please register via the Ticket Source link. Recordings of both talks with be made available on the website of the Centre for AI Fundamentals.
