A Topological Filter for Learning with Label Noise
This research provides a more robust method for training deep neural networks for practitioners and researchers dealing with datasets containing noisy labels, which is a common challenge in real-world applications.
This paper addresses the problem of noisy labels in deep neural networks by proposing a new method that filters label noise. The method leverages the spatial behavior and high-order topological information of data in the latent representational space to collect clean data, outperforming state-of-the-art methods and demonstrating robustness across various noise types and levels.
Noisy labels can impair the performance of deep neural networks. To tackle this problem, in this paper, we propose a new method for filtering label noise. Unlike most existing methods relying on the posterior probability of a noisy classifier, we focus on the much richer spatial behavior of data in the latent representational space. By leveraging the high-order topological information of data, we are able to collect most of the clean data and train a high-quality model. Theoretically we prove that this topological approach is guaranteed to collect the clean data with high probability. Empirical results show that our method outperforms the state-of-the-arts and is robust to a broad spectrum of noise types and levels.