Compressing Features for Learning With Noisy Labels

Yingyi Chen

doi:10.1109/TNNLS.2022.3186930

Compressing Features for Learning With Noisy Labels

2024-02-01

12

OA

AI

DOI：10.1109/TNNLS.2022.3186930

Original

Original request for help

Save

Abstract

En 中文

Supervised learning can be viewed as distilling relevant information from input data into feature representations. This process becomes difficult when supervision is noisy as the distilled information might not be relevant. In fact, recent research shows that networks can easily overfit all labels including those that are corrupted, and hence can hardly generalize to clean datasets. In this article, we focus on the problem of learning with noisy labels and introduce compression inductive bias to network architectures to alleviate this overfitting problem. More precisely, we revisit one classical regularization named Dropout and its variant Nested Dropout. Dropout can serve as a compression constraint for its feature dropping mechanism, while Nested Dropout further learns ordered feature representations with respect to feature importance. Moreover, the trained models with compression regularization are further combined with co-teaching for performance boost. Theoretically, we conduct bias variance decomposition of the objective function under compression regularization. We analyze it for both single model and co-teaching. This decomposition provides three insights: 1) it shows that overfitting is indeed an issue in learning with noisy labels; 2) through an information bottleneck formulation, it explains why the proposed feature compression helps in combating label noise; and 3) it gives explanations on the performance boost brought by incorporating compression regularization into co-teaching. Experiments show that our simple approach can have comparable or even better performance than the state-of-the-art methods on benchmarks with real-world label noise including Clothing1M and ANIMAL-10N. Our implementation is available at https://yingyichen-cyy.github.io/ CompressFeatNoisyLabels/.

Keywords:

Noise measurement

Training

Principal component analysis

Kernel

Deep learning

Biological system modeling

Benchmark testing

Bias variance decomposition

compression

deep learning

information sorting

label noise

Compressing Features for Learning With Noisy Labels

Abstract

Journal

Organization