Calibrating deep neural networks using focal loss

Conference item

Miscalibration -- a mismatch between a model's confidence and its correctness -- of Deep Neural Networks (DNNs) makes their predictions hard to rely on. Ideally, we want networks to be accurate, calibrated and confident. We show that, as opposed to the standard cross-entropy loss, focal loss (Lin et al., 2017) allows us to learn models that are already very well calibrated. When combined with temperature scaling, whilst preserving accuracy, it yields state-of-the-art calibrated models. We provide a thorough analysis of the factors causing miscalibration, and use the insights we glean from this to justify the empirically excellent performance of focal loss. To facilitate the use of focal loss in practice, we also provide a principled approach to automatically select the hyperparameter involved in the loss function. We perform extensive experiments on a variety of computer vision and NLP datasets, and with a wide variety of network architectures, and show that our approach achieves state-of-the-art calibration without compromising on accuracy in almost all cases. Code is available at https://github.com/torrvision/focal_calibration.

Authors: Mukhoti, J; Kulharia, V; Sanyal, A; Golodetz, S; Torr, PHS

• Publication status: Published
• Peer review status: Peer reviewed
• Publisher: Curran Associates
• Host title: Advances in Neural Information Processing Systems 33: 34th Conference on Neural Information Processing Systems (NeurIPS 2020)
• Volume: 33
• Pages: 15288-15299
• Series: Advances in Neural Information Processing Systems
• Publication date: 2020-12-06
• Acceptance date: 2020-09-25
• Event title: Thirty-fourth Conference on Neural Information Processing Systems (NeurIPS 2020)
• Event series: Neural Information Processing Systems
• Event location: Virtual event
• Event website: https://nips.cc/
• Event start date: 2020-12-06
• Event end date: 2020-12-12
• ISBN: 9781713829546
• Language: English
• Keywords: FFR