Summary of Noisy Label Processing For Classification: a Survey, by Mengting Li et al.
Noisy Label Processing for Classification: A Survey
by Mengting Li, Chuang Zhu
First submitted to arxiv on: 5 Apr 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper discusses the significance of combating noisy labels in computer vision tasks, particularly classification tasks. Noisy labels arise from the tedious process of manual annotation, where annotators may make mistakes. This can lead to incorrect image labels, which DNNs can easily fit, causing damage to model training. The survey reviews different deep learning approaches for noisy label combating and explores various noise patterns. It also proposes an algorithm to generate a synthetic label noise pattern guided by real-world data. The algorithm is tested on the CIFAR-10N dataset to create a new benchmark for evaluating noise-robust methods. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper talks about a big problem in computer vision: when people try to teach computers to recognize pictures, they make mistakes and give the wrong labels. This can cause the computer to learn incorrectly. The authors of this survey review ways that computer scientists are trying to fix this problem by using special kinds of algorithms. They also propose a new way to create fake noise patterns that mimic real-world mistakes. The goal is to help computers learn better and more accurately. |
Keywords
» Artificial intelligence » Classification » Deep learning
