Summary of Generalized Group Data Attribution, by Dan Ley et al.
Generalized Group Data Attribution
by Dan Ley, Suraj Srinivas, Shichang Zhang, Gili Rusak, Himabindu Lakkaraju
First submitted to arxiv on: 13 Oct 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
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 Generalized Group Data Attribution (GGDA) framework is introduced to simplify data attribution methods for large-scale machine learning models. Existing methods are often computationally intensive, but GGDA attributes influence to groups of training points instead of individual ones, providing a trade-off between efficiency and fidelity. The framework subsumes existing attribution methods like Influence Functions, TracIn, and TRAK, allowing users to optimize based on their needs. Empirical results demonstrate speedups of up to 10x-50x over standard DA methods while maintaining effectiveness in applications such as dataset pruning and noisy label identification. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary GGDA is a new way to figure out how individual pieces of training data affect machine learning models. Right now, these methods are too slow for big models, but GGDA makes them faster by grouping together similar pieces of data. This means you can use these methods on bigger models without it taking forever. It’s like getting more bang for your buck! |
Keywords
» Artificial intelligence » Machine learning » Pruning
