A Quantum-Classical Hybrid Solution for Deep Anomaly Detection

Abstract

Machine learning (ML) has achieved remarkable success in a wide range of applications. In recent ML research, deep anomaly detection (AD) has been a hot topic with the aim of discriminating among anomalous data with deep neural networks (DNNs). Notably, image AD is one of the most representative tasks in current deep AD research. ML’s interaction with quantum computing is giving rise to a heated topic named quantum machine learning (QML), which enjoys great prospects according to recent academic research. This paper attempts to address the image AD problem in a deep manner with a novel QML solution. Specifically, we design a quantum-classical hybrid DNN (QHDNN) that aims to learn directly from normal raw images to train a normality model and then exclude images that do not conform to this model as anomalies during its inference. To enable the QHDNN to perform satisfactorily in deep image AD, we explore multiple quantum layer architectures and design a VQC-based QHDNN solution. Extensive experiments were conducted on commonly used benchmarks to test the proposed QML solution, whose results demonstrate the feasibility of addressing deep image AD with QML. Importantly, the experimental results show that our quantum-classical hybrid solution can even yield superior performance to that of its classical counterpart when they share the same number of learnable parameters.

Quantum machine learning approach for classification: case studies and implications

“…This study focuses on the benefits of using quantum resources for particular learning task. The use of QML for anomaly identification in network traffic analysis was examined by Wang et al [22]. They put forth a framework for anomaly detection that was influenced by quantum theory and made use of a hybrid classical-quantum model and quantum feature space representation.…”

Diminishing Empirical Risk Minimization for Unsupervised Anomaly Detection

“…This endows ViT with the ability to locate subtle defects, an ability that is particularly important for precise anomaly detection tasks [110]. 3) For the practicability, as the modern foundational model, ViT can be effortlessly adapted for multi-modal inputs, which is endowed with versatility and potential in handling diverse AD applications in future works, i.e., multi-modal 3D AD [111], [112], zero-shot AD [40], [41], [113]. However, plain ViT has never been explored in AD because there seems to be a consensus among researchers that multi-resolution features are necessary to model accurate anomaly location, so most current AD methods would introduce a pyramidal network in Encoder [12], [21], [34] or Decoder [22], [34] to obtain multiresolution features.…”