| name | quantum-autoencoder-general-anomaly |
| description | Quantum convolutional autoencoder (QCAE) for reconstruction-based anomaly detection using QCNN architectures - semi-supervised training on normal samples with reconstruction error as anomaly score. |
| category | quantum |
| created | 2026-07-06T00:00:00.000Z |
| source | arXiv:2607.02135 |
Quantum Convolutional Autoencoders for Anomaly Detection
Source
arXiv:2607.02135 - "Quantum Convolutional Autoencoders for Reconstruction-Based Anomaly Detection" by Donovan Slabbert, Francesco Petruccione (2026-07-02)
Overview
Quantum convolutional neural networks (QCNNs) adapted into quantum autoencoder (QAE) framework for reconstruction-based anomaly detection. Models trained semi-supervised on normal samples with reconstruction error as anomaly score.
Core Methodology
-
QCNN-to-QAE Architecture: Adapt quantum convolutional neural network architecture into autoencoder framework. Hierarchical representation of quantum information enables efficient compression.
-
Two Architecture Variants: Models differ in treatment of latent information:
- Architecture A: Direct latent encoding
- Architecture B: Modified latent space handling
-
Semi-Supervised Training: Train on normal samples only to reconstruct feature-extracted and dimensionally reduced time-series data.
-
Reconstruction Error Scoring: Use reconstruction error as anomaly score - anomalies resist compression, producing higher reconstruction errors.
Key Findings
- QCNNs provide efficient parameterization for anomaly detection
- Hierarchical quantum representations capture data structure effectively
- Semi-supervised approach works well with limited normal data
- Two architecture variants offer different tradeoffs
- Applicable to scientific data analysis across domains
Applications
- Anomaly detection in time-series data
- Scientific data analysis
- Quantum machine learning pipelines
- Semi-supervised classification tasks
- Dimensionality reduction for anomaly detection
Trigger Words
quantum convolutional autoencoder, anomaly detection, QCNN, semi-supervised, reconstruction error, time-series analysis, quantum machine learning
Activation
When:
- Building anomaly detection systems with quantum circuits
- Working with quantum convolutional architectures
- Designing semi-supervised learning pipelines
- Analyzing time-series data for anomalies
- Exploring quantum advantage in unsupervised learning
