Summary of Attention Versus Contrastive Learning Of Tabular Data — a Data-centric Benchmarking, by Shourav B. Rabbani et al.
Attention versus Contrastive Learning of Tabular Data – A Data-centric Benchmarking
by Shourav B. Rabbani, Ivan V. Medri, Manar D. Samad
First submitted to arxiv on: 8 Jan 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 This research investigates the performance gap between deep learning and traditional machine learning methods on tabular data, highlighting the need for data-centric treatment and benchmarking. By evaluating state-of-the-art attention and contrastive learning methods on 28 tabular datasets against traditional deep and machine learning baselines, the study demonstrates that no single best learning method exists for all tabular data sets. The results show that combining between-sample and between-feature attentions conquers traditional ML on most datasets but fails on high-dimensional data, where contrastive learning takes a lead. A hybrid attention-contrastive learning strategy mostly wins on hard-to-classify datasets, while traditional methods are often superior on easy-to-classify datasets. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Deep learning has not improved significantly against traditional machine learning for tabular data. This paper tests advanced deep models on 28 datasets and finds that no single method works best for all. It combines attention and contrastive learning to beat traditional ML, but this combination fails on high-dimensional data where contrastive learning is better. The results show when to use each approach. |
Keywords
* Artificial intelligence * Attention * Deep learning * Machine learning
