Summary of Feature-space Semantic Invariance: Enhanced Ood Detection For Open-set Domain Generalization, by Haoliang Wang et al.
Feature-Space Semantic Invariance: Enhanced OOD Detection for Open-Set Domain Generalization
by Haoliang Wang, Chen Zhao, Feng Chen
First submitted to arxiv on: 11 Nov 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- 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 This paper tackles the challenge of open-set domain generalization, where a model must generalize across unseen domains and detect unknown classes not seen during training. Current approaches often tackle these issues separately, limiting their practical applicability. The authors propose a unified framework for open-set domain generalization by introducing Feature-space Semantic Invariance (FSI). FSI maintains semantic consistency across different domains within the feature space, enabling more accurate detection of out-of-distribution instances in unseen domains. Additionally, the authors adopt a generative model to produce synthetic data with novel domain styles or class labels, enhancing model robustness. Experimental results show that their method improves AUROC by 9.1% to 18.9% on ColoredMNIST while also increasing in-distribution classification accuracy. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper solves a big problem: training a model that can work with new things it’s never seen before. Right now, most models are only good at recognizing things they’ve been trained on. But what if you want to recognize something completely new? That’s where open-set domain generalization comes in. The authors of this paper have developed a way to make models better at handling new situations by making sure the features (or characteristics) of things stay consistent, even when the situation changes. They also use fake data to help their model get better at recognizing things it hasn’t seen before. So far, their method has been shown to be 9-18% better at recognizing unknown things and up to 20% better at recognizing things that are similar but not exactly the same. |
Keywords
» Artificial intelligence » Classification » Domain generalization » Generative model » Synthetic data
