Tiny but Trusted: Efficient Vision-Language Reasoning for Time-Series Anomaly Detection
For practitioners needing interpretable anomaly detection in time series, this work provides a more accurate and grounded VLM-based approach, though it is incremental as it applies existing fine-tuning techniques to a new domain.
The authors constructed VisAnomBench, a benchmark with anomaly explanations, and fine-tuned a parameter-efficient VLM (VisAnomReasoner) for time-series anomaly detection, achieving at least 21.23 and 23.87 percentage point improvements in precision and F1 on VisAnomBench, and 9.57 and 13.39 on TSB-AD-U.
Recent advances in Vision-Language Models (VLMs) have achieved impressive performance across many tasks, yet prior studies report unsatisfactory performance when applying large language or multimodal models to finding abnormal patterns in sequential data. Public anomaly detection benchmarks typically provide interval annotations but not natural-language rationales, making it difficult to fine-tune VLMs to produce grounded, interpretable decisions. To address this gap, we construct VisAnomBench, a curated benchmark built from public time-series datasets and augmented with high-quality anomaly explanations selected from multiple large VLMs using fine-grained, task-specific rewards. Through fine-tuning on this benchmark, we develop VisAnomReasoner, a parameter-efficient VLM for time-series anomaly detection. Experimental results on VisAnomBench show that VisAnomReasoner achieves more accurate anomaly localization and consistently outperforms all baselines, with improvements of at least 21.23 and 23.87 percentage points in precision and F1, respectively. Additional experiments on the TSB-AD-U benchmark demonstrate strong cross-benchmark generalization, with VisAnomReasoner improving precision and F1 by 9.57 and 13.39 percentage points, respectively.