The perils of being unhinged: On the accuracy of classifiers minimizing a noise-robust convex loss
Not on the map. Why?
Abstract (arXiv)
Van Rooyen et al. introduced a notion of convex loss functions being robust to random classification noise, and established that the "unhinged" loss function is robust in this sense. In this note we study the accuracy of binary classifiers obtained by minimizing the unhinged loss, and observe that even for simple linearly separable data distributions, minimizing the unhinged loss may only yield a binary classifier with accuracy no better than random guessing.
Key facts. Cited by 2 papers in the map (data through 7 Oct 2026). Builds on 8 of its 33 references that are in the map. First posted on arXiv 8 Dec 2021.
Builds on
8 of 33 references.
References · 8
- 2018Generalized Cross Entropy Loss for Training Deep Neural Networks with Noisy Labelscited by 994
- 2016Making Deep Neural Networks Robust to Label Noise: a Loss Correction Approachcited by 663
- 2017The Implicit Bias of Gradient Descent on Separable Datacited by 649
- 2017Robust Loss Functions under Label Noise for Deep Neural Networkscited by 441
- 2015Learning with Symmetric Label Noise: The Importance of Being Unhingedcited by 120
- 2014Making Risk Minimization Tolerant to Label Noisecited by 54
- 2019On Symmetric Losses for Learning from Corrupted Labelscited by 53
- 2013Margins, Shrinkage, and Boostingcited by 35
Built on by
Citing papers per year: 2023: 1; 2024: 0; 2025: 0; 2026: 1.
2 papers.