Summary of Seqnas: Neural Architecture Search For Event Sequence Classification, by Igor Udovichenko et al.
SeqNAS: Neural Architecture Search for Event Sequence Classification
by Igor Udovichenko, Egor Shvetsov, Denis Divitsky, Dmitry Osin, Ilya Trofimov, Anatoly Glushenko, Ivan Sukharev, Dmitry Berestenev, Evgeny Burnaev
First submitted to arxiv on: 6 Jan 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 proposed SeqNAS algorithm addresses the limitation of previous Neural Architecture Search (NAS) methods, which were only applied to image, text, or time series domains. SeqNAS is specifically designed for event sequence classification, a common task in industries such as churn prediction, customer segmentation, fraud detection, and fault diagnosis. The algorithm develops a simple yet expressive search space using building blocks like multihead self-attention, convolutions, and recurrent cells. Bayesian Optimization is employed to perform the search, while an ensemble of teacher models augments knowledge distillation. As a result, SeqNAS surpasses state-of-the-art NAS methods and popular architectures for sequence classification. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary SeqNAS is a new way to find the best solution for event sequences, like predicting customer behavior or detecting fraud. Event sequences are special because they have different types of data (like numbers and categories) and irregular timestamps. Previous solutions only worked for other types of data, like images or text. SeqNAS can be used in many industries, including finance, healthcare, and manufacturing. |
Keywords
* Artificial intelligence * Classification * Knowledge distillation * Optimization * Self attention * Time series
