Talk@Simlab : Building trust in neural networks for energy modelling

Speaker:  Prof. Antoine Lesage-Landry

13 Gennaio 2026 | 11:15
DEIB, Sala Carlo Erba (Ed. 7)
Piazza Leonardo da Vinci, 32 

Contatti:  Prof. Giambattista Gruosso

Sommario

On January 13^th, 2026, at 11:15 am the seminar on "Talk@Simlab : Building trust in neural networks for energy modelling" will take place in DEIB Carlo Erba Room (Building 7).

While neural networks have empirically shown their performance as non-linear predictors, they suffer from low interpretability, limited out-of-the-box performance guarantees, complex training procedures, and high susceptibility to data corruption and other adversarial attacks. Combined altogether, this results in a low acceptability for application in critical sectors like energy systems. In this context, we propose Wasserstein distributionally robust shallow convex neural networks (WaDiRo-SCNNs) to provide reliable nonlinear predictions when subject to adverse and corrupted datasets. Our training procedure is conservative by design, has low stochasticity, is solvable with open-source solvers, and is scalable to large industrial deployments. Our approach aims to make neural networks safer for critical applications, such as in the energy sector. Finally, we numerically demonstrate the performance of our model on a synthetic experiment and a real-world power system application, i.e., the prediction of non-residential buildings' hourly energy consumption in the context of virtual power plants.

Biografia

Antoine Lesage-Landry is an Associate Professor of Electrical Engineering at Polytechnique Montréal, a Member of the Group for Research in Decision Analysis (GERAD) and an Academic Associate Member of Mila — Québec’s AI Institute. He received the B.Eng. degree in Engineering Physics from Polytechnique Montréal, in 2015 and the Ph.D. degree in Electrical Engineering from the University of Toronto in 2019. From 2019 to 2020, I was a Postdoctoral Scholar in the Energy & Resources Group at the University of California, Berkeley. His research activities focus on the mathematical analysis of decision-making in electric power systems and wireless communications. His areas of expertise include operations research, machine learning, and power engineering.