Summary of Zero-shot Generalist Graph Anomaly Detection with Unified Neighborhood Prompts, by Chaoxi Niu et al.
Zero-shot Generalist Graph Anomaly Detection with Unified Neighborhood Prompts
by Chaoxi Niu, Hezhe Qiao, Changlu Chen, Ling Chen, Guansong Pang
First submitted to arxiv on: 18 Oct 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 proposed UNPrompt approach is a novel zero-shot generalist graph anomaly detection method that trains a single model on a single graph dataset and then effectively detects anomalies in other graph datasets without retraining or fine-tuning. This is achieved through two main modules: one module aligns node attributes across different graphs via coordinate-wise normalization, while another module learns generalized neighborhood prompts to support latent node attribute predictability as an anomaly score. UNPrompt significantly outperforms competing methods under both the generalist and one-model-for-one-dataset settings. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary UNPrompt is a new way to find unusual patterns in graphs. Usually, we need to train a separate model for each graph dataset, but this can be hard or even impossible in real-world scenarios. UNPrompt trains just one model on a single graph dataset and then uses it to detect anomalies in other datasets without needing any more training. This is possible because the approach learns generalized patterns in graphs that work across different datasets. |
Keywords
» Artificial intelligence » Anomaly detection » Fine tuning » Zero shot
