Summary of Tagfog: Textual Anchor Guidance and Fake Outlier Generation For Visual Out-of-distribution Detection, by Jiankang Chen and Tong Zhang and Wei-shi Zheng and Ruixuan Wang
TagFog: Textual Anchor Guidance and Fake Outlier Generation for Visual Out-of-Distribution Detection
by Jiankang Chen, Tong Zhang, Wei-Shi Zheng, Ruixuan Wang
First submitted to arxiv on: 22 Nov 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Machine Learning (cs.LG)
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 learning framework for out-of-distribution (OOD) detection leverages Jigsaw-based fake OOD data and rich semantic embeddings, or “anchors,” from ChatGPT’s description of in-distribution (ID) knowledge to guide the training of an image encoder. This framework can be combined with existing post-hoc approaches to improve OOD detection performance. Extensive empirical evaluations on multiple benchmarks demonstrate that incorporating textual representation of ID and fake OOD knowledge enables state-of-the-art results. The code is available at https://github.com/Cverchen/TagFog. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary A new way to detect when images are from outside the normal range has been developed. This method uses fake images created using a technique called Jigsaw, along with special “anchors” that describe what’s in these fake images. The goal is to help train an image recognition system to do better at detecting when it sees something it hasn’t seen before. By combining this new approach with others already being used, the best results yet have been achieved on multiple tests. |
Keywords
* Artificial intelligence * Encoder
