Energy-Based Models for Anomaly Detection: A Manifold Diffusion Recovery Approach

TL;DR

This paper introduces MPDR, a novel energy-based model training method that exploits low-dimensional data structures for improved anomaly detection across multiple data modalities.

Contribution

The paper proposes a new training algorithm for EBMs that uses manifold perturbations and MCMC-generated negative samples to enhance anomaly detection capabilities.

Findings

MPDR achieves high accuracy in diverse anomaly detection tasks.

The method effectively models data boundaries and detects out-of-distribution samples.

Experimental results demonstrate superior performance over existing approaches.

Abstract

We present a new method of training energy-based models (EBMs) for anomaly detection that leverages low-dimensional structures within data. The proposed algorithm, Manifold Projection-Diffusion Recovery (MPDR), first perturbs a data point along a low-dimensional manifold that approximates the training dataset. Then, EBM is trained to maximize the probability of recovering the original data. The training involves the generation of negative samples via MCMC, as in conventional EBM training, but from a different distribution concentrated near the manifold. The resulting near-manifold negative samples are highly informative, reflecting relevant modes of variation in data. An energy function of MPDR effectively learns accurate boundaries of the training data distribution and excels at detecting out-of-distribution samples. Experimental results show that MPDR exhibits strong performance across various anomaly detection tasks involving diverse data types, such as images, vectors, and acoustic signals.