The Perils of Optimizing Learned Reward Functions: Low Training Error Does Not Guarantee Low Regret
Fluri, L Lang, L Abate, A Forré, P Krueger, D Skalse, J Proceedings of Machine Learning Research volume 267 17306-17377 (01 Jan 2025)
Fri, 06 Mar 2026
16:00
L1

We are all different: Modeling key individual differences in physiological systems

Anita Layton
(University of Waterloo)
Abstract
Mathematical models of whole-body dynamics have advanced our understanding of human integrative systems that regulate physiological processes such as metabolism, temperature, and blood pressure. For most of these whole-body models, baseline parameters describe a 35-year-old young adult man who weighs 70 kg. As such, even among adults those models may not accurately represent half of the population (women), the older population, and those who weigh significantly more than 70 kg. Indeed, sex, age, and weight are known modulators of physiological function. To more accurately simulate a person who does not look like that “baseline person,” or to explain the mechanisms that yield the observed sex or age differences, these factors should be incorporated into mathematical models of physiological systems. Another key modulator is the time of day, because most physiological processes are regulated by the circadian clocks. Thus, ideally, mathematical models of integrative physiological systems should be specific to either a man or woman, of a certain age and weight, and a given time of day. A major goal of our research program is to build models specific to different subpopulations, and conduct model simulations to unravel the functional impacts of individual differences.


 

Detection of anomalous spatio-temporal patterns of app traffic in response to catastrophic events
Pedreschi, N Medina, S Lambiotte, R LaRock, T Babul, S Sahasrabuddhe, R
Quantifying the Spatial and Demographic Scales of Segregation
Sahasrabuddhe, R Lambiotte, R (28 Nov 2025)
Manipulating Collective Opinion through Social Network Intervention
Hata, S Lambiotte, R Nakao, H Kobayashi, R (16 Nov 2025)
Complex-Weighted Convolutional Networks: Provable Expressiveness via Complex Diffusion
Amado, C Schwarz, T Tian, Y Lambiotte, R (17 Nov 2025)
Embedding networks with the random walk first return time distribution
Thapar, V Lambiotte, R Cantwell, G (03 Dec 2025)
Thu, 11 Jun 2026

14:00 - 15:00
Lecture Room 3

Optimization Algorithms for Bilevel Learning with Applications to Imaging

Dr Lindon Roberts
(Melbourne University)
Abstract

Dr Lindon Roberts will talk about: 'Optimization Algorithms for Bilevel Learning with Applications to Imaging'

Many imaging problems, such as denoising or inpainting, can be expressed as variational regularization problems. These are optimization problems for which many suitable algorithms exist. We consider the problem of learning suitable regularizers for imaging problems from example (training) data, which can be formulated as a large-scale bilevel optimization problem. 

In this talk, I will introduce new deterministic and stochastic algorithms for bilevel optimization, which require no or minimal hyperparameter tuning while retaining convergence guarantees. 

This is joint work with Mohammad Sadegh Salehi and Matthias Ehrhardt (University of Bath), and Subhadip Mukherjee (IIT Kharagpur).

 

 

Subscribe to