Keynote Speakers

Title: AutoML From Raw Spatio-Temporal Observations to Decisions

Modern sensing technologies have provided the possibility of sensing the world in a way that has not been possible before, generating massive spatio-temporal data sources. How can we use such data to understand and even change the complex world around us for the better? In this talk, I will discuss unique machine learning challenges in transforming such data into actionable decisions. These challenges call for automated solutions to address various problems, from filling the gaps in the data to filling the gaps in the knowledge acquired from data alone. I will present a few examples of such problems and automated solutions to address them.

Mitra Baratchi is assistant professor of artificial intelligence at Leiden University, where she leads the Spatio-temporal data Analysis and Reasoning (STAR) and co-leads of the Automated Design of Algorithms research group. Her research interests lie in spatio-temporal, time-series, and mobility data modelling. She strongly focuses on developing algorithms for wearable sensors data, Earth observations and other open spatio-temporal data sources. Specifically, she explores the design of algorithms that can automatically handle all necessary data processing tasks from the point of data collection to high-level modelling, extraction of information, and effective decision-making from such data. Her research targets applications in a broad range of urban, environmental, and industrial domains, for which she has collaborated, notably with the European Space Agency, Netherlands Institute for Space Research, Honda Research Institute, various municipalities, and researchers in other scientific disciplines.

Irina Rish is a Full Professor in the Computer Science and Operations Research Department at the Université de Montréal (UdeM) and a core faculty member of MILA – Quebec AI Institute, where she leads the Autonomous AI Lab. Dr. Rish holds a Canada Excellence Research Chair (CERC) in Autonomous AI and a Canadian Institute for Advanced Research (CIFAR) Canada AI Chair. Her extensive research career spans multiple AI domains, from automated reasoning and probabilistic inference in graphical models, to machine learning, sparse modeling, and neuroscience-inspired AI. Her current research endeavors concentrate on continual learning, out-of-distribution generalization, and robustness of AI systems, as well as understanding neural scaling laws and emergent behaviors, w.r.t. both capabilities and alignment, in large-scale foundation models – a vital stride towards achieving maximally beneficial Artificial General Intelligence (AGI). She teaches courses on AI scaling and alignment, and runs Neural Scaling & Alignment workshop series. Dr. Rish is a recipient of the INCITE 2023 and other compute grants by the US Department of Energy, and leads several projects on Scalable Foundation Models on Summit & Frontier supercomputers at the Oak Ridge Leadership Computing Facility, focusing on developing open-source large-scale AI models. She is also a co-founder and the Chief Science Officer of nolano.ai, a company focused on training, compression and fast inference in large-scale AI models.

Title: Learning the Dynamics & the Dynamics of Learning

We start by looking at how structure imposition in hierarchical models can transform dynamical system learning. Not all problems are so readily solved by imposing structure priors. We will look at some of the concepts behind understanding the loss-surfaces and dynamics of the learning process. Understanding the complexities of deep-model loss-surfaces leads to better learning procedures & improved generalisation performance, with some interesting insights along the way.

Steve Roberts is Professor of Machine Learning at the University of Oxford. He is a Fellow of the Royal Academy of Engineering, co-leads Oxford’s Eric & Wendy Schmidt AI in Science Programme, is director of the Oxford ELLIS unit and is co-founder of the Oxford AI company Mind Foundry. Steve’s interests lie in the theory, methodology and application of machine learning to real-world problem domains. His current research includes applications in astrophysics, climate & environment, ecology, finance and engineering as well as a range of theoretical and methodological problems.

Title: New Frontiers and Opportunities for AutoML

AutoML began by optimizing hyperparameters such as learning rate and regularization. Quickly, however, AutoML embraced larger pieces of the ML pipeline and hyperparameters eventually included things like the choice of learning algorithm and pre- and post-processing steps such as feature coding, feature selection, missing value imputation, model calibration and neural net architecture. Now LLMs and transformers have created new opportunities for AutoML. And methods such as causal learning and high-accuracy glass-box learning suggest that future AutoML may need to focus less on accuracy and more on correctness.

Rich Caruana is a senior principal researcher at Microsoft Research. Before joining Microsoft, Rich was on the faculty in the Computer Science Department at Cornell University, at UCLA’s Medical School, and at CMU’s Center for Learning and Discovery. Rich’s Ph.D. is from Carnegie Mellon University, where he worked with Tom Mitchell and Herb Simon. His thesis on Multi-Task Learning helped create interest in a new subfield of machine learning called Transfer Learning. Rich received an NSF CAREER Award in 2004 (for Meta Clustering), best paper awards in 2005 (with Alex Niculescu-Mizil), 2007 (with Daria Sorokina), and 2014 (with Todd Kulesza, Saleema Amershi, Danyel Fisher, and Denis Charles), co-chaired KDD in 2007 (with Xindong Wu), and serves as area chair for NIPS, ICML, and KDD. His current research focus is on learning for medical decision making, transparent modeling, deep learning, and computational ecology.

Title: Machine Learning in the Optimization and Discovery Loop

Many problems in science and engineering, from discovering novel molecules to tuning machine learning systems, can be viewed as estimating and optimizing an unknown function that is accessible only through noisy experiments. The field of Bayesian optimization seeks to address these challenges by quantifying uncertainty about the unknown objective and utilizing the uncertainty to navigate the exploration—exploitation dilemma. In this talk, I will present recent work motivated by key challenges in complex applications such as protein design, robotics and AutoML. In particular, I will discuss meta-learning of probabilistic models from related tasks and simulations, directing exploration in combinatorial search spaces via reinforcement learning and exploiting causal structure to improve both sample efficiency and interpretability.

Andreas Krause is a Professor of Computer Science at ETH Zurich, where he leads the Learning & Adaptive Systems Group. He also serves as Academic Co-Director of the Swiss Data Science Center and Chair of the ETH AI Center, and co-founded the ETH spin-off LatticeFlow. Before that he was an Assistant Professor of Computer Science at Caltech. He received his Ph.D. in Computer Science from Carnegie Mellon University (2008) and his Diplom in Computer Science and Mathematics from the Technical University of Munich, Germany (2004). He is a Max Planck Fellow at the Max Planck Institute for Intelligent Systems, ACM Fellow, ELLIS Fellow, a Microsoft Research Faculty Fellow and a Kavli Frontiers Fellow of the US National Academy of Sciences. He received the Rössler Prize, ERC Starting Investigator and ERC Consolidator grants, the German Pattern Recognition Award, an NSF CAREER award as well as the ETH Golden Owl teaching award. His research on machine learning and adaptive systems has received awards at several premier conferences and journals, including the ACM SIGKDD Test of Time award 2019 and the ICML Test of Time award 2020. Andreas Krause served as Program Co-Chair for ICML 2018 and General Chair for ICML 2023 and serves as Action Editor for the Journal of Machine Learning Research. In 2023, he was appointed to the United Nations’ High-level Advisory Body on AI.