AnomSeer: Reinforcing Multimodal LLMs to Reason for Time-Series Anomaly Detection

Time-series anomaly detection (TSAD) with multimodal large language models (MLLMs) is an emerging area, yet a persistent challenge remains: MLLMs rely on coarse time-series heuristics but struggle with multi-dimensional, detailed reasoning, which is vital for understanding complex time-series data. We present AnomSeer to address this by reinforcing the model to ground its reasoning in precise, structural details of time series, unifying anomaly classification, localization, and explanation. At its core, an expert chain-of-thought trace is generated to provide verifiable, fine-grained reasoning from classical analyses (e.g., statistical measures, frequency transforms). Building on this, we propose a novel time-series grounded policy optimization (TimerPO) that incorporates two additional components beyond standard reinforcement learning: a time-series grounded advantage based on optimal transport and an orthogonal projection to ensure this auxiliary granular signal does not interfere with the primary detection objective. Across diverse anomaly scenarios, AnomSeer, with Qwen2.5-VL-3B/7B-Instruct, outperforms larger commercial baselines in classification and localization accuracy, particularly on point- and frequency-driven exceptions. Moreover, it produces plausible reasoning traces that support its conclusions.

Many real-world systems, such as hospitals, factories, and websites, produce streams of numbers that change over time, and unusual patterns in these streams can signal important problems. Today’s AI systems can sometimes spot these unusual patterns in charts, but they often rely on rough visual impressions and may not clearly explain what changed or where it happened. We introduce AnomSeer, an AI method that learns to examine time-based data more carefully by looking for detailed changes in shape, rhythm, sudden spikes, and local patterns. It can identify the kind of unusual behavior, locate the affected time period, and explain its decision in plain language. This makes AI-based monitoring more accurate and easier to understand in real-world settings.