Summary of Mitigating Label Noise on Graph Via Topological Sample Selection, by Yuhao Wu et al.
Mitigating Label Noise on Graph via Topological Sample Selection
by Yuhao Wu, Jiangchao Yao, Xiaobo Xia, Jun Yu, Ruxin Wang, Bo Han, Tongliang Liu
First submitted to arxiv on: 4 Mar 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
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 proposes a novel approach called Topological Sample Selection (TSS) to enhance the effectiveness of graph neural networks (GNNs) when dealing with noisy labels in real-world graph data. The existing benchmarks are carefully annotated, but previous studies on sample selection have focused on i.i.d data and do not account for non-iid graph data or GNNs. The TSS method leverages topological information to improve the informative sample selection process in a graph, minimizing an upper bound of the expected risk under the target clean distribution. Compared to state-of-the-art baselines, experimental results demonstrate the superiority of the proposed method. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper solves a problem with graph neural networks (GNNs) when dealing with noisy labels in real-world data. Normally, GNNs work well, but not when there’s noise in the labels. People have tried to fix this by picking the best samples, but it hasn’t worked well because those methods don’t take into account the special structure of graphs. The new method, called Topological Sample Selection (TSS), fixes this problem by using information about how the data is connected. This makes GNNs work much better with noisy labels. |
