Summary of Tabmda: Tabular Manifold Data Augmentation For Any Classifier Using Transformers with In-context Subsetting, by Andrei Margeloiu et al.
TabMDA: Tabular Manifold Data Augmentation for Any Classifier using Transformers with In-context Subsetting
by Andrei Margeloiu, Adrián Bazaga, Nikola Simidjievski, Pietro Liò, Mateja Jamnik
First submitted to arxiv on: 3 Jun 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 introduces TabMDA, a novel method for manifold data augmentation on tabular data. This approach utilizes pre-trained in-context models, such as TabPFN, to map the data into an embedding space. By performing label-invariant transformations and encoding the data multiple times with varied contexts, TabMDA explores the learned embedding space of the underlying models, effectively enlarging the training dataset. This method is training-free, making it applicable to any classifier. The authors evaluate TabMDA on five standard classifiers and observe significant performance improvements across various tabular datasets. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Tabular data is important in many areas, but it can be hard to get a lot of it. When we try to use machine learning models with this type of data, they often don’t perform well because there’s not enough information. A common way to make machine learning models better is by adding fake data that looks like the real data. This usually works well for pictures and language, but not as much for tabular data. The authors came up with a new way to do this called TabMDA. It uses a special kind of model that’s already learned about the data, and it makes many copies of the data in different ways. This helps the machine learning models learn more from the small amount of real data we have. The authors tested TabMDA on several types of tabular data and found that it made the models work much better. |
Keywords
» Artificial intelligence » Data augmentation » Embedding space » Machine learning
