Reproduce “Enhancing Time-Series Anomaly Detection by Integrating Spectral-Residual Bottom-Up Attention with Reservoir Computing” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Hayato Nihei, Sou Nobukawa, Yusuke Sakemi, Kazuyuki Aihara
The authors introduce SR-RC and Multi-SR-RC for time-series anomaly detection, pairing a training-free spectral-residual saliency map with a recurrent reservoir whose output weights are trained on labeled anomalies. Multi-SR-RC also receives the series. On synthetic benchmarks, models outperform larger conventional reservoirs for some anomaly types; on Yahoo S5 Task 13, Multi-SR-RC reaches mean F1 1.000 among six series.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.