# Training-Free Human-in-the-Loop Anomaly Detection via Memory Bank Correction

arXiv: https://arxiv.org/abs/2608.17775 · HTML: https://mlsanity.com/paper/2608.17775/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 23 of its 41 references that are in the map.
- First posted on arXiv 18 Aug 2026.

## Builds on (most cited first, 23 of 23)

- [Wide Residual Networks](https://mlsanity.com/paper/1605.07146/) (2016)
- [Towards Total Recall in Industrial Anomaly Detection](https://mlsanity.com/paper/2106.08265/) (2021)
- [PaDiM: a Patch Distribution Modeling Framework for Anomaly Detection and Localization](https://mlsanity.com/paper/2011.08785/) (2020)
- [Unsupervised Anomaly Detection with Generative Adversarial Networks to Guide Marker Discovery](https://mlsanity.com/paper/1703.05921/) (2017)
- [WinCLIP: Zero-/Few-Shot Anomaly Classification and Segmentation](https://mlsanity.com/paper/2303.14814/) (2023)
- [Sub-Image Anomaly Detection with Deep Pyramid Correspondences](https://mlsanity.com/paper/2005.02357/) (2020)
- [SimpleNet: A Simple Network for Image Anomaly Detection and Localization](https://mlsanity.com/paper/2303.15140/) (2023)
- [AnomalyGPT: Detecting Industrial Anomalies Using Large Vision-Language Models](https://mlsanity.com/paper/2308.15366/) (2023)
- [Student-Teacher Feature Pyramid Matching for Anomaly Detection](https://mlsanity.com/paper/2103.04257/) (2021)
- [Registration based Few-Shot Anomaly Detection](https://mlsanity.com/paper/2207.07361/) (2022)
- [Pushing the Limits of Fewshot Anomaly Detection in Industry Vision: Graphcore](https://mlsanity.com/paper/2301.12082/) (2023)
- [Unveiling the Anomalies in an Ever-Changing World: A Benchmark for Pixel-Level Anomaly Detection in Continual Learning](https://mlsanity.com/paper/2403.15463/) (2024)
- [Effectiveness of Tree-based Ensembles for Anomaly Discovery: Insights, Batch and Streaming Active Learning](https://mlsanity.com/paper/1901.08930/) (2019)
- [Taming False Positives in Out-of-Distribution Detection with Human Feedback](https://mlsanity.com/paper/2404.16954/) (2024)
- [Active Anomaly Detection via Ensembles](https://mlsanity.com/paper/1809.06477/) (2018)
- [Little Help Makes a Big Difference: Leveraging Active Learning to Improve Unsupervised Time Series Anomaly Detection](https://mlsanity.com/paper/2201.10323/) (2022)
- [A Reliable Framework for Human-in-the-Loop Anomaly Detection in Time Series](https://mlsanity.com/paper/2405.03234/) (2024)
- [CADIC: Continual Anomaly Detection Based on Incremental Coreset](https://mlsanity.com/paper/2511.08634/) (2025)
- [Human in the AI loop via xAI and Active Learning for Visual Inspection](https://mlsanity.com/paper/2307.05508/) (2023)
- [ASAT: Adaptive Scoring and Thresholding with Human Feedback for Robust Out-of-Distribution Detection](https://mlsanity.com/paper/2505.02299/) (2025)
- [Leveraging Intermediate Representations of Time Series Foundation Models for Anomaly Detection](https://mlsanity.com/paper/2509.12650/) (2025)
- [MECAD: A multi-expert architecture for continual anomaly detection](https://mlsanity.com/paper/2512.15323/) (2025)
- [Normality-Preserving Continual Industrial Anomaly Detection via Orthogonal LoRA Banks](https://mlsanity.com/paper/2606.02042/) (2026)

## Abstract (arXiv)

Anomaly detectors are hardest to deploy exactly where training data is scarcest: a newly commissioned production line has a handful of verified "golden" samples and no machine-learning engineer on the factory floor. We present a training-free human-in-the-loop framework in which a domain expert corrects a PatchCore detector by direct memory bank editing: no retraining, no gradients, no original training data. A false-positive correction inserts the reviewed image's normal patches through a self-calibrating novelty gate admitting only those beyond the median pool-normal nearest-neighbour distance. From a bank built on only ten golden samples, operator corrections close a median 66% of the gap to an uncorrected fully trained bank (mean 80%, raised by three categories that overshoot parity), significantly improving 12 of 15 MVTec AD categories and harming none: ten samples plus corrections outperform hundreds of samples without them. On already-trained banks the headroom is smaller and concentrated where the bank undersamples normal appearance (gated: toothbrush +0.10, metal nut +0.09, zipper +0.05, screw +0.05), and no category except grid is significantly harmed. Evaluation uses a held-out protocol (20 splits per category, Holm-corrected Wilcoxon), because corrected images entering the bank inflate naive evaluation toward AUROC 1.0 by memorisation. Passive and active querying are statistically indistinguishable; a matched-label-budget control attributes gains to deployment-time label production at 43% of exhaustive-review cost; a defect-memory extension fails decisively. Feedback is simulated from ground truth; live expert trials, where mislabelling is costliest on small banks, remain future work.