Quorum: Zero-Training Unsupervised Anomaly Detection using Quantum Autoencoders

TL;DR

Quorum tackles unsupervised anomaly detection on near-term quantum hardware by eliminating training. It leverages amplitude encoding, random quantum transformations, and SWAP tests to detect anomalies as deviations from the norm without gradient-based optimization. The work introduces a training-free, ensemble-based framework with bucketing and feature selection, showing strong performance and noise resilience across multiple datasets. The results suggest practical implications for deploying quantum anomaly detectors in real-world, unlabeled-data settings.

Abstract

Detecting mission-critical anomalous events and data is a crucial challenge across various industries, including finance, healthcare, and energy. Quantum computing has recently emerged as a powerful tool for tackling several machine learning tasks, but training quantum machine learning models remains challenging, particularly due to the difficulty of gradient calculation. The challenge is even greater for anomaly detection, where unsupervised learning methods are essential to ensure practical applicability. To address these issues, we propose Quorum, the first quantum anomaly detection framework designed for unsupervised learning that operates without requiring any training.

Figures (10)

• Figure 1: Visual representation of the steps taken by Quorum to detect the anomalies in a given dataset using quantum computing.
• Figure 2: Quorum's use of SWAP test to determine the similarity between the compressed and original data sample.
• Figure 3: Quorum's bucketing procedure for distributing the subsampled data samples across different ensemble groups.
• Figure 4: Quorum subsamples features from the input space.
• Figure 5: The ansatz leveraged by Quorum for its autoencoder to establish correlations across input features.