Summary of Mixstyle-entropy: Domain Generalization with Causal Intervention and Perturbation, by Luyao Tang et al.
Mixstyle-Entropy: Domain Generalization with Causal Intervention and Perturbation
by Luyao Tang, Yuxuan Yuan, Chaoqi Chen, Xinghao Ding, Yue Huang
First submitted to arxiv on: 7 Aug 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Computer Vision and Pattern Recognition (cs.CV); Methodology (stat.ME)
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 This paper tackles the issue of deep neural networks’ poor performance when applied to unseen environments. Domain Generalization (DG) aims to solve this problem by learning representations independent of domain-specific information. Existing approaches focus on training objectives that extract shared features from source data, but this may compromise robustness in real-world scenarios. The proposed framework, InPer, incorporates causal intervention during training and perturbation during testing to enhance model generalization. It uses entropy-based causal intervention (EnIn) to refine variable selection and a novel metric, homeostatic score (HoPer), to construct a prototype classifier at test time. Experimental results across multiple cross-domain tasks confirm the effectiveness of InPer. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper helps deep learning models work better when they’re used in new situations. Right now, these models can be very good at recognizing things like pictures or speech, but they often don’t generalize well to new environments. This means they might not perform as well on new data that’s a bit different from what they were trained on. Domain Generalization (DG) is an area of research that tries to solve this problem by teaching models to ignore information specific to one environment and focus on general features. The authors of this paper propose a new approach called InPer, which uses ideas from causality to improve model performance. They test their approach on several real-world tasks and show it works well. |
Keywords
» Artificial intelligence » Deep learning » Domain generalization » Generalization
