Summary of Beyond the Known: Novel Class Discovery For Open-world Graph Learning, by Yucheng Jin and Yun Xiong and Juncheng Fang and Xixi Wu and Dongxiao He and Xing Jia and Bingchen Zhao and Philip Yu
Beyond the Known: Novel Class Discovery for Open-world Graph Learning
by Yucheng Jin, Yun Xiong, Juncheng Fang, Xixi Wu, Dongxiao He, Xing Jia, Bingchen Zhao, Philip Yu
First submitted to arxiv on: 29 Mar 2024
Categories
- Main: Machine Learning (cs.LG)
- 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 proposes a novel approach to node classification on graphs, tackling the challenge of discovering new classes that emerge on unlabeled nodes in real-world scenarios. The authors argue that existing methods struggle when dealing with correlations between known and novel classes, which makes their representations indistinguishable. To address this issue, they introduce Open-world gRAph neuraL network (ORAL), a semi-supervised learning method that detects correlations through prototypical learning, eliminates inter-class correlations using attention networks, and generates pseudo-labels by ensembling label estimations from multiple stacked networks. The authors demonstrate the effectiveness of ORAL on several benchmark datasets. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper tries to solve a big problem in computers: how do we teach machines to recognize new types of things on graphs when we can’t give them labels? Right now, if we don’t have labels for some nodes, our current methods get stuck. The authors came up with a new way to handle this called Open-world gRAph neuraL network (ORAL). It works by finding patterns between different types of things, ignoring the connections that make it hard to tell them apart, and then guessing what type something is based on how similar it is to labeled things. They tested ORAL on some real datasets and it worked really well. |
Keywords
* Artificial intelligence * Attention * Classification * Graph neural network * Semi supervised
