Summary of Domain Expansion and Boundary Growth For Open-set Single-source Domain Generalization, by Pengkun Jiao et al.
Domain Expansion and Boundary Growth for Open-Set Single-Source Domain Generalization
by Pengkun Jiao, Na Zhao, Jingjing Chen, Yu-Gang Jiang
First submitted to arxiv on: 5 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 A novel learning approach is proposed to tackle the challenge of open-set single-source domain generalization, where a model must be learned from a limited source domain and generalized to unknown target domains with both domain shifts and label shifts. The method, based on domain expansion and boundary growth, synthesizes new samples by applying background suppression and style augmentation to the source data. This expanded source dataset is then used to train a model that distills consistent knowledge about the source domain. Additionally, edge maps are employed as an additional modality for training multi-binary classifiers, allowing for boundary growth across classes and improved unknown class recognition during open-set generalization. Experimental results demonstrate significant improvements and state-of-the-art performance on several cross-domain image classification datasets. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper solves a big problem in artificial intelligence called domain generalization. It’s like trying to teach a robot to recognize animals in different environments, even if it only learned from one type of environment before. The authors developed a new way to make the robot learn more by making fake images that are similar to the real ones. They also found a way to make the robot better at recognizing new things by using maps that show where edges are on an image. This helps the robot understand what is inside and outside a picture, which makes it better at recognizing things it hasn’t seen before. |
Keywords
» Artificial intelligence » Domain generalization » Generalization » Image classification
