Summary of First-order Manifold Data Augmentation For Regression Learning, by Ilya Kaufman and Omri Azencot
First-Order Manifold Data Augmentation for Regression Learning
by Ilya Kaufman, Omri Azencot
First submitted to arxiv on: 16 Jun 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
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 Data augmentation (DA) techniques generate synthetic samples tailored to specific domains, such as image rotations or time series transformations. In contrast, domain-independent approaches like mixup are versatile and applicable across various data modalities. While DA is well-studied for classification tasks, its effect on regression problems has received less attention. To bridge this gap, we introduce FOMA: a new domain-independent DA method that samples new examples from the tangent planes of the train distribution. Our approach aligns with neural networks’ tendency to capture dominant features. We evaluate FOMA on in-distribution generalization and out-of-distribution robustness benchmarks, showing it improves several neural architectures’ generalization. Interestingly, strong baselines based on mixup are less effective compared to our approach. Our code is publicly available at https://github.com/azencot-group/FOMA. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper explores how to make machine learning models better at predicting numbers and patterns. Right now, most models focus on specific types of data, like pictures or sounds. But what if we want a model that can work with different kinds of data? The researchers created a new way to mix up the data, called FOMA, which makes the model more accurate and robust. They tested it with several different models and found that it worked really well. This is important because it means we can use these models for even more things in the future. |
Keywords
» Artificial intelligence » Attention » Classification » Data augmentation » Generalization » Machine learning » Regression » Time series
