CRITS: Convolutional Rectifier for Interpretable Time Series Classification
Alejandro Kuratomi ✉, Zed Lee, Guilherme Dinis Chaliane Junior, Tony Lindgren, Diego García Pérez ✉ · 2025
Read CRITS: Convolutional Rectifier for Interpretable Time Series Classification by Alejandro Kuratomi ✉, Zed Lee, Guilherme Dinis Chaliane Junior, Tony Lindgren, Diego García Pérez ✉ (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-12042.txt.
Description
Kuratomi and co-authors present a classifier, published in 2025, for data that unfold over time, designed to explain its own choices. Time series classifiers are used in health, finance, and industry, but most act as black boxes. Common explanation tools paint rough heat maps that blur the exact stretch of signal that drove the call. Others jiggle the input many times, which is slow. The team's model builds the explanation into its structure.
A layer of pattern filters scans the series, and a simple scoring network weighs what they find. The stretches that fire the winning filters are the explanation, shown at full detail in the original signal. Because the explanation is intrinsic, it costs no extra runs. The authors test the model on standard time series tasks. They show it classifies well while pointing plainly at the evidence it used.
Catalog record
| CREATOR | Alejandro Kuratomi ✉, Zed Lee, Guilherme Dinis Chaliane Junior, Tony Lindgren, Diego García Pérez ✉ |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 34 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-12042