Summary of Anytime Neural Architecture Search on Tabular Data, by Naili Xing et al.
Anytime Neural Architecture Search on Tabular Data
by Naili Xing, Shaofeng Cai, Zhaojing Luo, Beng Chin Ooi, Jian Pei
First submitted to arxiv on: 15 Mar 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 The paper introduces ATLAS, an anytime Neural Architecture Search (NAS) approach tailored for tabular data analysis. This efficient and responsive method returns current optimal architectures within a given time budget while progressively enhancing architecture quality with increased budget allocation. ATLAS combines the strengths of training-free and training-based architecture evaluation using a novel two-phase filtering-and-refinement optimization scheme with joint optimization. The filter phase employs a zero-cost proxy to estimate candidate architecture performance, narrowing down promising candidates for refinement in the second phase. A fixed-budget search algorithm schedules training to accurately identify the optimal architecture. Experimental evaluations demonstrate ATLAS’s ability to obtain good-performing architectures within any time budget and improve upon existing NAS approaches, reducing search time by up to 82.75x. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary ATLAS is a new way for computers to find the best design for analyzing tables of data. Right now, people have to manually design these designs, which takes a lot of time. The goal of ATLAS is to make this process faster and better. It does this by using two different methods to evaluate different designs and then choosing the best one. This helps the computer find a good design quickly, even if it doesn’t have all day. In tests, ATLAS was able to find good designs much faster than other methods. |
Keywords
* Artificial intelligence * Optimization
