Summary of Semi Supervised Heterogeneous Domain Adaptation Via Disentanglement and Pseudo-labelling, by Cassio F. Dantas (evergreen et al.
Semi Supervised Heterogeneous Domain Adaptation via Disentanglement and Pseudo-Labelling
by Cassio F. Dantas, Raffaele Gaetano, Dino Ienco
First submitted to arxiv on: 20 Jun 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- 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 This paper proposes SHeDD, a semi-supervised heterogeneous domain adaptation framework for learning a target domain classifier from labelled and unlabelled data from diverse sources. The setting, known as Semi-Supervised Heterogeneous Domain Adaptation (SSHDA), poses an even greater challenge due to the distribution shift caused by modality heterogeneity across domains. SHeDD disentangles domain-invariant representations from domain-specific information using an end-to-end neural framework and incorporates augmentation-based consistency regularization for improved generalization. Experimental results on remote sensing benchmarks demonstrate SHeDD’s superiority over baseline and state-of-the-art approaches. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary SHeDD is a new way to help machines learn from different types of data, even if they come from different sources or have different characteristics. This is important because it can be hard for machines to adapt to new situations when the data they’ve learned from looks very different from what they’re trying to predict now. The authors came up with a special kind of AI framework that can do this adaptation by separating out things that are important for learning (like patterns in the data) and things that aren’t (like differences between how the data was collected). They also added some extra tricks to make their AI model more robust and able to generalize well. The results show that SHeDD performs better than other methods on tasks like classifying images from different sensors or acquisition modes. |
Keywords
» Artificial intelligence » Domain adaptation » Generalization » Regularization » Semi supervised
