Neural Architecture Search of Sample Reweighting Networks for Complex Distribution Shift
2026-06-22 • Machine Learning
Machine LearningArtificial Intelligence
AI summaryⓘ
The authors studied a method called Meta-Weight-Net (MW-Net) that helps adjust how much focus a learning model gives to each example when data has problems like noisy labels or uneven class sizes. They found that MW-Net works well for one problem at a time but struggles when both problems happen together. To fix this, they used a technique called neural architecture search to find the best network design and input features for MW-Net. Their tests on modified image datasets showed that this approach helps MW-Net handle multiple data issues better.
sample reweightingdistribution shiftlabel noiseclass imbalanceMeta-Weight-Netneural architecture searchtree-structured Parzen estimatorCIFAR-10CIFAR-100classification loss
Authors
Keisuke Sugawara, Kento Uchida, Shinichi Shirakawa
Abstract
Sample reweighting is a major approach to addressing distribution shifts, such as label noise and class imbalance. Meta-Weight-Net (MW-Net) is a promising sample reweighting network that computes weights based on classification loss. Although MW-Net improves prediction performance under a single type of distribution shift using a simple neural network, its performance degrades when facing both label noise and class imbalance, where it is hard to determine appropriate weights solely from classification loss and using a simple network. In this study, we introduce neural architecture search to MW-Net to mitigate such performance degradation. Using the tree-structured Parzen estimator, we explore the optimal number of hidden layers and nodes and select the most suitable intermediate layer in the classification model to serve as the input for MW-Net. Experimental results on the CIFAR-10 and CIFAR-100 datasets that were modified to include both label noise and class imbalance demonstrate the effectiveness of neural architecture search for MW-Net.