ICAD-LLM: One-for-All Anomaly Detection via In-Context Learning with Large Language Models
Zhongyuan Wu, Jingyuan Wang, Zexuan Cheng, Yilong Zhou, Weizhi Wang, Juhua Pu, Chao Li, Changqing Ma · 2025
Read ICAD-LLM: One-for-All Anomaly Detection via In-Context Learning with Large Language Models by Zhongyuan Wu, Jingyuan Wang, Zexuan Cheng, Yilong Zhou, Weizhi Wang, Juhua Pu, Chao Li, Changqing Ma (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-2512-01672.txt.
Description
Zhongyuan Wu and co-authors build one anomaly detector for many data types with language models. Modern IT systems emit time series, logs, and tables at once. Each type once needed its own detector. Fast change in those systems makes upkeep hard.
Their ICAD-LLM uses in-context learning to adapt to each type. Few examples in the prompt set the task. One model then covers diverse streams. A single adaptive detector could cut the cost of guarding large, mixed IT systems against faults and attacks. Logs, metrics, and tables differ wildly in shape, so one learner for all three is a strong test of in-context skill itself.
Catalog record
| CREATOR | Zhongyuan Wu, Jingyuan Wang, Zexuan Cheng, Yilong Zhou, Weizhi Wang, Juhua Pu, Chao Li, Changqing Ma |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 45 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2512-01672