Summary of One-shot Domain Incremental Learning, by Yasushi Esaki and Satoshi Koide and Takuro Kutsuna
One-Shot Domain Incremental Learning
by Yasushi Esaki, Satoshi Koide, Takuro Kutsuna
First submitted to arxiv on: 25 Mar 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
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 study explores the challenges of domain incremental learning (DIL) in deep neural networks, focusing on the extreme case where only one sample from a new domain is available. Existing DIL methods are found to be ineffective in this scenario, and the analysis reveals that the issue stems from statistics in batch normalization layers. To address this problem, the researchers propose a technique that tackles these statistics and demonstrate its effectiveness through experiments on open datasets. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary In this study, scientists investigate how machines learn when new information is added. They discovered that existing methods don’t work well when they only have one example from the new data. By looking closely at why this happens, they found that a key problem lies in how some layers process statistics. To fix this issue, the researchers developed a solution and tested it on public datasets to show its success. |
Keywords
* Artificial intelligence * Batch normalization
