Reproduce “Quantum Autoencoder for Multivariate Time Series Anomaly Detection” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Kilian Tscharke, Maximilian Wendlinger, Afrae Ahouzi, Pallavi Bhardwaj, Kaweh Amoi-Taleghani, Michael Schrödl-Baumann, Pascal Debus
A quantum autoencoder (QAE) developed by Fraunhofer AISEC and SAP SE detects anomalies in multivariate time series data from enterprise systems, achieving performance comparable to medium-sized classical autoencoders with significantly fewer trainable parameters.
Key insights
- Quantum autoencoders (QAEs) can achieve competitive anomaly detection performance on multivariate time series data while requiring significantly fewer trainable parameters than classical autoencoders.
- The QAE effectively captures complex inter-variable dependencies through quantum entanglement, which is crucial for detecting subtle anomalies in multivariate time series.
- A modified loss function, incorporating a term for 'trash qubits', improves the trainability and performance of the quantum autoencoder.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.