黑料福利网

Benjamin Quost, Professor at Universit茅 de Technologie de Compi猫gne (UTC, France) is a guest of Cassio de Campos, Full Professor of research group Uncertainty in AI, Department M&CS, 黑料福利网.

Title  |  Learning from poor information: a glance behind, a glimpse ahead 

The recent trends in machine learning involve complex models learnt from huge amounts of data. However, in a wide range of applications, the information at hand may be scarce, noisy, or imprecise. Processing such poor data remains a challenge; an alternative then consists to learn models with nice robustness properties, and (whenever possible) to involve humans during training or when making decisions. In this talk, we take a glance at several such cases where such poor data can be processed to produce robust or cautious models; we also discuss how uncertainty can be leveraged for this purpose. Eventually, we glimpse at several (open) research directions aiming at producing trustful AI systems, making connections with the topics of fairness and explainability.

Program
15:30 - 16:15 Lecture in Neuron 0.262 (doors open at 15:15)
16:15 - 16:30 Q&A
16:30 - 17:00 Drinks

is required and free of charge.