LLMs for Automated Algorithm Design
Lars Kotthoff, University of St Andrews, United Kingdom, United Kingdom
Available Soon
Hervé Panetto, University of Lorraine, France, France
Towards Explainable AI for Bipolar Disorder Monitoring
Gabriella Casalino, Università degli Studi di Bari, Italy, Italy
Brief Bio
Lars is a professor at the University of St Andrews, where he holds the Johann
and Gaynor Rupert Chair in Artificial Intelligence, and a visiting professorship
at Sorbonne Université. Previously, he has held chairs at the University of
Wyoming and postdoctoral positions at the University of British Columbia and
University College Cork. Lars holds a PhD from the University of St Andrews. His
research focuses on automated algorithm design in diverse application domains,
with automated machine learning as one of the most prominent areas, and
applications of AI in other disciplines, most prominently materials science.
Abstract
Automatically designing high-performing algorithms in diverse problem domains
has been an active research area for a long time. Instead of laboriously
designing algorithms manually, the vision is to efficiently and effectively
create new algorithms that are tailored to a particular application, leverage
the structure of a problem better than existing algorithms, or work under more
restrictive constraints. The involvement of human experts is reduced to
providing guidance, rather than doing all the heavy lifting.
In this keynote, I will argue that LLMs are the missing piece to achieve this
vision. They are surprisingly good at generating code, and combined with
established and new techniques in automated algorithm design, have demonstrated
that they can beat state-of-the-art expert-designed algorithms in several
application domains. While LLMs are part of the solution, challenges still
remain and interesting research questions are waiting to be tackled.
Brief Bio
Dr. Hervé Panetto is a Professor of Enterprise Information Systems at University of Lorraine, TELECOM Nancy. He teaches Information Systems modelling and development, and conducts research at CRAN (Research Centre for Automatic Control), Joint Research Unit with CNRS where he is managing a research project on the use of ontology for formalising models related to the interoperability of production systems, and mainly their enterprise information systems. He has been elected member of the Academia Europaea in 2018. He has been elected, in 2020, Chairman of the IFAC French National Member Organization (NMO).
He received his PhD in production engineering in 1991. He has strong experience in information systems modelling, semantics modelling and discovery, and database development. His research field is based on information systems modelling for enterprise applications and processes interoperability, with applications in enterprise modelling, manufacturing processes modelling, furniture data modelling. He is working in ERP and MES integration from a Business to manufacturing perspective. He is expert at AFNOR (French National standardisation body), CEN TC310 and ISO TC184/SC4 and SC5. He participated in many European projects including IMS FP5-IST Smart-fm project (awarded by IMS) and the FP6 INTEROP NoE (Interoperability Research for Networked Enterprises Applications and Software). He is serving as expert-evaluator for the European Commission, FNR, AERES and ANR in the domain of ICT. He was visiting Professor in 2013-2015 in the frame of a Science Without Borders PVE project with PUC Parana, Brazil and full visiting Professor in 2016 at the UTFPR, Curitiba, Brazil. He is editor or guest editor of books and special issues of international journals. He is author or co-author of more than 150 papers in the field of Automation Engineering, Enterprise Modelling and Enterprise systems integration and interoperability. After being Chair of the IFAC Technical Committee 5.3 “Enterprise Integration and Networking” from 2008 to 2014. He is Chair of the IFAC Coordinating Committee 5 on “Manufacturing and Logistics Systems” since 2014. He received the IFAC France Award 2013, the INCOSE 2015 Outstanding Service Award and the IFAC 2017 Outstanding Service Award. He is co-organiser of the yearly OTM/IFAC/IFIP EI2N workshop on “Enterprise Integration, Interoperability and Networking”. He is General Co-chair of the OTM Federated conferences. He is member of the Editorial Board of the Annual Reviews in Control, Computers In Industry, the International Journal of Computer Integrated Manufacturing, the International Journal on Universal Computer Science, the scientific journal Facta Universitatis, series Mechanical Engineering, and an Associate Editor of the international Journal of Intelligent Manufacturing (JIM), Springer, the Enterprise Information Systems (EIS) journal, Taylor & Francis, the IEEE Internet of Things Journal, the Journal of Industrial Information Integration (JIII), Elsevier, and the Journal SN Computer Science (SNCS), Springer Nature.
Brief Bio
Gabriella Casalino is currently a postdoctoral research fellow at the CILab laboratory of the department of Informatics, University of Bari, working on fuzzy logic based information filtering systems.
Her research activity is focused on the development of intelligent data analysis techniques. Her work has spanned a diverse set of topics, at the intersection of different fields. Topics in which she has produced original contributions include: image analysis, educational data mining, text mining, e-health, bioinformatics and signal processing.
Gabriella Casalino has got the Ph.D. in Computer Science at the Doctoral School in Computer Science of the Department of Informatics, at University of Bari "A. Moro" in 2015. She defended the thesis "Non-negative factorization methods for extracting semantically relevant features in Intelligent Data Analysis" under the supervision of Prof. Corrado Mencar and Prof. Nicoletta Del Buono. She was awarded by the Italian Ministry of Education, University and Research (M.I.U.R.) with a grant covering the three-year period of her Ph.D. Between February 2014 and June 2014, she spent five months in Mons, Belgium, for a research internship at the Department of Mathematics and Operational Research, Facultè polytechnique, Universitè de Mons, supervised by Prof. Nicolas Gillis.
In 2008 she got a B.Sc. in Computer Science from the University of Bari, and in 2011 she got a M.Sc. in Computer Science from the same University.
She is active in the computer science community as reviewers for international journals and conferences.
Abstract
Bipolar disorder is a complex mental illness characterized by pronounced mood fluctuations, often accompanied by distinctive changes in speech patterns. Clinicians intuitively rely on these vocal cues during assessments. With today’s ability to collect speech data continuously through smartphones, mobile technologies offer a promising pathway for early detection and real-time monitoring of mood episodes.
Despite this potential, smartphone-based acoustic approaches remain underused in clinical practice. A major barrier is the lack of standardized machine learning methods, transparent evaluation metrics, and clinically interpretable explanations that clarify how acoustic markers relate to symptoms and mood dynamics.
In this talk, I will present my research on developing explainable AI methods for bipolar disorder monitoring, with a focus on interpretable, clinically meaningful models grounded in fuzzy logic and explainable acoustic biomarkers. The goal is to bridge the gap between data-driven systems and clinical reasoning, moving toward AI-assisted mental health tools that are trustworthy, actionable, and aligned with psychiatric practice.