Memorizing Normality to Detect Anomaly: Memory-augmented Deep Autoencoder for Unsupervised Anomaly Detection

Abstract

Deep autoencoder has been extensively used for anomaly detection. Training on the normal data, the autoencoder is expected to produce higher reconstruction error for the abnormal inputs than the normal ones, which is adopted as a criterion for identifying anomalies. However, this assumption does not always hold in practice. It has been observed that sometimes the autoencoder "generalizes" so well that it can also reconstruct anomalies well, leading to the miss detection of anomalies. To mitigate this drawback for autoencoder based anomaly detector, we propose to augment the autoencoder with a memory module and develop an improved autoencoder called memory-augmented autoencoder, i.e. MemAE. Given an input, MemAE firstly obtains the encoding from the encoder and then uses it as a query to retrieve the most relevant memory items for reconstruction. At the training stage, the memory contents are updated and are encouraged to represent the prototypical elements of the normal data. At the test stage, the learned memory will be fixed, and the reconstruction is obtained from a few selected memory records of the normal data. The reconstruction will thus tend to be close to a normal sample. Thus the reconstructed errors on anomalies will be strengthened for anomaly detection. MemAE is free of assumptions on the data type and thus general to be applied to different tasks. Experiments on various datasets prove the excellent generalization and high effectiveness of the proposed MemAE.

Accepted to appear at ICCV 2019

References in corpus (9)

• Adam: A Method for Stochastic Optimization
• Conditional Image Generation with PixelCNN Decoders
• Deep Anomaly Detection Using Geometric Transformations
• One-shot Learning with Memory-Augmented Neural Networks
• Deep Structured Energy Based Models for Anomaly Detection
• Anomaly Detection using One-Class Neural Networks
• Learning Deep Representations of Appearance and Motion for Anomalous Event Detection
• Memorization Precedes Generation: Learning Unsupervised GANs with Memory Networks
• Learning to Generate with Memory

Cited by in corpus (19)

• DFR: Deep Feature Reconstruction for Unsupervised Anomaly Segmentation
• A Survey on Deep Learning Techniques for Video Anomaly Detection
• Encoding Structure-Texture Relation with P-Net for Anomaly Detection in Retinal Images
• Image/Video Deep Anomaly Detection: A Survey
• MIST: Multiple Instance Self-Training Framework for Video Anomaly Detection
• Anomaly Detection with Domain Adaptation
• Identifying the Defective: Detecting Damaged Grains for Cereal Appearance Inspection
• Innovations Autoencoder and its Application in One-class Anomalous Sequence Detection
• Few-Shot Anomaly Detection for Polyp Frames from Colonoscopy
• Video Anomaly Detection By The Duality Of Normality-Granted Optical Flow
• Unsupervised Dual Adversarial Learning for Anomaly Detection in Colonoscopy Video Frames
• Exemplar-Based Open-Set Panoptic Segmentation Network
• Neural Memory Plasticity for Anomaly Detection
• Unsupervised Learning for Industrial Defect Detection: A Case Study on Shearographic Data
• Evaluating the Use of Reconstruction Error for Novelty Localization
• Autoencoding Under Normalization Constraints
• ODDObjects: A Framework for Multiclass Unsupervised Anomaly Detection on Masked Objects
• Sub-clusters of Normal Data for Anomaly Detection
• Reconstruct Anomaly to Normal: Adversarial Learned and Latent Vector-constrained Autoencoder for Time-series Anomaly Detection