Summary of Beyond Model Adaptation at Test Time: a Survey, by Zehao Xiao et al.
Beyond Model Adaptation at Test Time: A Survey
by Zehao Xiao, Cees G. M. Snoek
First submitted to arxiv on: 6 Nov 2024
Categories
- Main: Machine Learning (cs.LG)
- 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 The paper presents a comprehensive review of test-time adaptation, a machine learning approach that combines domain adaptation and domain generalization to address distribution shifts between training and testing data. The authors categorize over 400 recent papers into five categories based on what component is adjusted for test-time adaptation: model, inference, normalization, sample, or prompt. The review provides detailed analysis of each category, discussing preparation and adaptation settings, and offering insights into the effective deployment for evaluating distribution shifts and real-world applications in understanding images, video, 3D, and beyond. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Test-time adaptation is a new way to make machine learning models work better when they encounter data that’s different from what they were trained on. Most machine learning models are good at recognizing things like pictures of cats or dogs, but they struggle when the pictures are taken in a different lighting or angle. Test-time adaptation tries to fix this by adjusting the model during testing so it can understand new types of data. The authors of this paper reviewed over 400 papers on test-time adaptation and grouped them into five categories based on what part of the model is adjusted. They also discussed how these methods work in different settings, like evaluating distribution shifts or understanding images, video, and 3D. |
Keywords
» Artificial intelligence » Domain adaptation » Domain generalization » Inference » Machine learning » Prompt
