# TrustMAE: A Noise-Resilient Defect Classification Framework using Memory-Augmented Auto-Encoders with Trust Regions

arXiv: https://arxiv.org/abs/2012.14629 · HTML: https://mlsanity.com/paper/2012.14629/

## Key facts

- Cited by 21 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 29 Dec 2020.

## Built on by (most cited first, 21 of 21)

- [Deep Industrial Image Anomaly Detection: A Survey](https://mlsanity.com/paper/2301.11514/) (2023)
- [Deep Learning for Unsupervised Anomaly Localization in Industrial Images: A Survey](https://mlsanity.com/paper/2207.10298/) (2022)
- [IM-IAD: Industrial Image Anomaly Detection Benchmark in Manufacturing](https://mlsanity.com/paper/2301.13359/) (2023)
- [Diversity-Measurable Anomaly Detection](https://mlsanity.com/paper/2303.05047/) (2023)
- [DiAD: A Diffusion-based Framework for Multi-class Anomaly Detection](https://mlsanity.com/paper/2312.06607/) (2023)
- [FAPM: Fast Adaptive Patch Memory for Real-time Industrial Anomaly Detection](https://mlsanity.com/paper/2211.07381/) (2022)
- [LafitE: Latent Diffusion Model with Feature Editing for Unsupervised Multi-class Anomaly Detection](https://mlsanity.com/paper/2307.08059/) (2023)
- [AMI-Net: Adaptive Mask Inpainting Network for Industrial Anomaly Detection and Localization](https://mlsanity.com/paper/2412.11802/) (2024)
- [Normal Reference Attention and Defective Feature Perception Network for Surface Defect Detection](https://mlsanity.com/paper/2211.10060/) (2022)
- [IPAD: Industrial Process Anomaly Detection Dataset](https://mlsanity.com/paper/2404.15033/) (2024)
- [Domain-Generalized Textured Surface Anomaly Detection](https://mlsanity.com/paper/2203.12304/) (2022)
- [ALMRR: Anomaly Localization Mamba on Industrial Textured Surface with Feature Reconstruction and Refinement](https://mlsanity.com/paper/2407.17705/) (2024)
- [Clear Memory-Augmented Auto-Encoder for Surface Defect Detection](https://mlsanity.com/paper/2208.03879/) (2022)
- [PatchEAD: Unifying Industrial Visual Prompting Frameworks for Patch-Exclusive Anomaly Detection](https://mlsanity.com/paper/2509.25856/) (2025)
- [Restoring Noisy Demonstration for Imitation Learning With Diffusion Models](https://mlsanity.com/paper/2510.14467/) (2025)
- [URA-Net: Uncertainty-Integrated Anomaly Perception and Restoration Attention Network for Unsupervised Anomaly Detection](https://mlsanity.com/paper/2603.22840/) (2026)
- [SCL-VI: Self-supervised Context Learning for Visual Inspection of Industrial Defects](https://mlsanity.com/paper/2311.06504/) (2023)
- [A Feature Shuffling and Restoration Strategy for Universal Unsupervised Anomaly Detection](https://mlsanity.com/paper/2603.22861/) (2026)
- [Template-Based Feature Aggregation Network for Industrial Anomaly Detection](https://mlsanity.com/paper/2603.22874/) (2026)
- [DuoAD: Leveraging [CLS] Dual Characteristics for Training-Free Few-Shot Anomaly Detection](https://mlsanity.com/paper/2607.23924/) (2026)
- [Bayesian Tensor Autoencoder with Physics-informed Predictive Prior for Multi-dimensional Time Series Anomaly Detection](https://mlsanity.com/paper/2609.31157/) (2026)

## Abstract (arXiv)

In this paper, we propose a framework called TrustMAE to address the problem of product defect classification. Instead of relying on defective images that are difficult to collect and laborious to label, our framework can accept datasets with unlabeled images. Moreover, unlike most anomaly detection methods, our approach is robust against noises, or defective images, in the training dataset. Our framework uses a memory-augmented auto-encoder with a sparse memory addressing scheme to avoid over-generalizing the auto-encoder, and a novel trust-region memory updating scheme to keep the noises away from the memory slots. The result is a framework that can reconstruct defect-free images and identify the defective regions using a perceptual distance network. When compared against various state-of-the-art baselines, our approach performs competitively under noise-free MVTec datasets. More importantly, it remains effective at a noise level up to 40% while significantly outperforming other baselines.