Summary of Open-world Semi-supervised Learning For Node Classification, by Yanling Wang et al.
Open-World Semi-Supervised Learning for Node Classification
by Yanling Wang, Jing Zhang, Lingxi Zhang, Lixin Liu, Yuxiao Dong, Cuiping Li, Hong Chen, Hongzhi Yin
First submitted to arxiv on: 18 Mar 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Social and Information Networks (cs.SI)
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 proposed IMbalance-Aware method, OpenIMA, tackles open-world semi-supervised node classification in graphs, where nodes are classified into seen classes or novel classes. The issue arises when seen classes have smaller intra-class variances than novel classes due to imbalance, negatively impacting model performance. Pre-trained feature encoders can help alleviate this imbalance but require graph-specific training. OpenIMA addresses this challenge by proposing a scratch-based approach using contrastive learning with bias-reduced pseudo labels for open-world node classification. Experimental results on seven graph benchmarks demonstrate the effectiveness of OpenIMA. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Open-world semi-supervised learning is like teaching machines to recognize things they’ve never seen before, just as we learn new concepts without labeled examples. The problem is that these new things are harder to learn because they don’t have clear labels, making it challenging for models to perform well. One reason for this struggle is the imbalance between what’s familiar and what’s new. To overcome this, researchers developed a method called OpenIMA that starts from scratch instead of relying on pre-trained knowledge. The results show that OpenIMA can effectively classify nodes in graphs into seen or new categories. |
Keywords
* Artificial intelligence * Classification * Semi supervised
