On Label Noise in Image Classification
An Aleatoric Uncertainty Perspective
Time: Mon 2024-06-03 09.00
Location: F3 (Flodis), Lindstedsvägen 26 & 28
Video link: https://kth-se.zoom.us/w/61097277235
Language: English
Subject area: Computer Science
Doctoral student: Erik Englesson , Robotik, perception och lärande, RPL
Opponent: Professor Gustavo Carneiro, University of Surrey
Supervisor: Associate Professor Hossein Azizpour, Robotik, perception och lärande, RPL; Associate Professor Morteza Haghir Chehreghani, Chalmers University of Technology, Gothenburg, Sweden
QC 20240516
Abstract
Deep neural networks and large-scale datasets have revolutionized the field of machine learning. However, these large networks are susceptible to overfitting to label noise, resulting in generalization degradation. In response, the thesis closely examines the problem both from an empirical and theoretical perspective. We empirically analyse the input smoothness of networks as they overfit to label noise, and we theoretically explore the connection to aleatoric uncertainty. These analyses improve our understanding of the problem and have led to our novel methods aimed at enhancing robustness against label noise in classification.