{"slug":"anomaly-detection","label":"Anomaly Detection","item_count":3,"day_count":2,"source_count":3,"first_seen":"2026-08-24T16:36:17+00:00","last_updated":"2026-08-25T16:52:32+00:00","generated_at":"2026-09-01T20:11:18.457385+00:00","sources":["arxiv_cs_ai","arxiv_cs_lg","arxiv_llm_reliability"],"days":[{"date":"2026-08-24","items":[{"title":"RAD: Rule-Augmented Relational Anomaly Detection","url":"http://arxiv.org/abs/2608.23468v1","source":"arxiv_cs_lg","type":"paper","summary_1line":"Anomaly detection is often applied to data stored in relational databases, yet most existing methods require flattening multiple tables into a single feature matrix. This flattening can obscure entity identity, schema...","why_it_matters":"Matches feed focus: eval.","sid":"c1f5f7897cab6194","published":"2026-08-24T16:36:17+00:00","editor_note":"RAD proposes rule-augmented anomaly detection that operates directly on relational database structure instead of a flattened feature matrix."},{"title":"Robustness of Anomaly Detection Models for Industrial Control Systems under Training-Time Data Contamination","url":"http://arxiv.org/abs/2608.23547v1","source":"arxiv_llm_reliability","type":"paper","summary_1line":"Machine-learning-based anomaly detection is increasingly used in industrial control systems (ICS), yet most studies assume that detector training data is trustworthy. In practice, training data may be corrupted throug...","why_it_matters":"Matches feed focus: eval.","sid":"c6589f453fe3efd2","published":"2026-08-24T17:49:36+00:00","editor_note":"A study measures how training-time data contamination degrades industrial-control-system anomaly detectors."}]},{"date":"2026-08-25","items":[{"title":"Strictly Causal Streaming Video Anomaly Detection with a Theoretically-Grounded State-Space Core","url":"http://arxiv.org/abs/2608.24810v1","source":"arxiv_cs_ai","type":"paper","summary_1line":"Recent work has applied Mamba style state space models (SSMs) to video anomaly detection, yet existing approaches still rely on buffering clips or windows internally, lack a theoretical account of how temporal memory...","why_it_matters":"Matches feed focus: evaluation.","sid":"c8fa2c59f3b3c019","published":"2026-08-25T16:52:32+00:00","editor_note":"A strictly causal, state-space model targets real-time streaming video anomaly detection without internal clip buffering."}]}],"editorial":{"tldr":"Three papers published within two days push anomaly detection along separate, unconnected fronts: relational-database structure, industrial-control-system security, and real-time video. None builds on the others — this is a cluster of concurrent research, not one unfolding event.","stale":false,"whats_new":"The newest paper (Aug 25) proposes a causal, state-space model for streaming video anomaly detection that avoids buffering clips, unlike prior Mamba-style approaches.","why_it_matters":"If you're building anomaly detection into an agent or observability pipeline, these three papers map to distinct decision points: whether your data is relational (schema-aware methods), whether your training data can be adversarially poisoned (ICS robustness), and whether you need real-time, low-latency video inference (streaming state-space models).","take_for_builders":"Treat this as three independent leads, not one trend: check RAD if your anomaly signal lives in relational/warehouse data, check the ICS contamination study if your training pipeline ingests third-party or adversarial data, and check the streaming state-space model only if you need real-time video inference without clip buffering.","beats":[{"kicker":"RELATIONAL DATA","tone":"neutral","headline":"RAD argues anomaly detection over relational databases shouldn't flatten tables into one feature matrix","summary":"Flattening multiple tables loses entity identity and schema structure; RAD instead applies rule-augmented detection directly on the relational structure.","sids":["c1f5f7897cab6194"]},{"kicker":"SECURITY ROBUSTNESS","tone":"neutral","headline":"A study tests industrial-control-system anomaly detectors under training-time data contamination","summary":"Most ICS anomaly-detection research assumes clean training data; this paper measures what happens when that data is poisoned instead.","sids":["c6589f453fe3efd2"]},{"kicker":"STREAMING VIDEO","tone":"rising","headline":"A causal state-space model targets real-time video anomaly detection without clip buffering","summary":"Prior Mamba-style approaches to video anomaly detection still buffer clips or windows internally; this model claims a theoretically grounded, strictly causal alternative.","sids":["c8fa2c59f3b3c019"]}],"open_questions":["Do any of these three methods report benchmarks against a shared production anomaly-detection baseline, or only against prior academic work in their own subfield?","Does RAD's relational approach scale to the table counts and schema complexity found in real production data warehouses?","Has the ICS contamination study identified which specific defenses (if any) hold up under the poisoning attacks it tests?"],"generated_at":"2026-08-27T05:20:00+00:00"}}