Graph-Augmented LSTM for Forecasting Sparse Anomalies in Graph-Structured Time Series
Sneh Pillai · 2025
Read Graph-Augmented LSTM for Forecasting Sparse Anomalies in Graph-Structured Time Series by Sneh Pillai (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-2503-03729.txt.
Description
Sneh Pillai proposes adding graph structure to a forecasting model to catch rare anomalies in linked time series, published in 2025. Spotting anomalies is a key task in many fields. It gets harder when anomalies are sparse and the data come from many linked sensors or nodes, where each series affects the others.
Pillai builds the links among series directly into an LSTM forecasting model. Forecasts that miss badly can then flag events that a single-series view would miss. The method was tested on two benchmark data sets, a web anomaly set and a traffic sensor network. Against three standard baselines, the graph model reached higher precision and recall, with a clear gain in the combined F1 score.
Catalog record
| CREATOR | Sneh Pillai |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 27 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-03729