Searching to Exploit Memorization Effect in Learning from Corrupted Labels
This work addresses a hard problem in noisy-label learning for machine learning practitioners, offering an incremental improvement with a more efficient AutoML-based approach.
The paper tackles the problem of controlling sample selection in robust learning from noisy labels to exploit the memorization effect, proposing a novel Newton algorithm for efficient bi-level optimization and demonstrating superior performance and efficiency over state-of-the-art methods on benchmark datasets.
Sample selection approaches are popular in robust learning from noisy labels. However, how to properly control the selection process so that deep networks can benefit from the memorization effect is a hard problem. In this paper, motivated by the success of automated machine learning (AutoML), we model this issue as a function approximation problem. Specifically, we design a domain-specific search space based on general patterns of the memorization effect and propose a novel Newton algorithm to solve the bi-level optimization problem efficiently. We further provide theoretical analysis of the algorithm, which ensures a good approximation to critical points. Experiments are performed on benchmark data sets. Results demonstrate that the proposed method is much better than the state-of-the-art noisy-label-learning approaches, and also much more efficient than existing AutoML algorithms.