Agnostiq, Ontario, Canada
The task of identifying abnormal behavior in time series data is essential in many fields ranging from financial markets to biophysics. While traditional algorithms for time series anomaly detection (TAD) have proven to be effective, the advent of newly accessible quantum processing units (QPUs) presents an opportunity to explore a quantum approach to TAD. This talk will introduce a new TAD algorithm called Quantum Variational Rewinding (QVR)[1,2,3], where we map and learn time series processes by embedding classical data into quantum states and rewinding them to their learnt initial state – the process we colloquially call as rewinding with quantum time machine. This talk will also prove as a gentile introduction of quantum circuits, quantum machine learning and end with discussion of the possible advantages of why one would consider QVR over traditional classical TAD algorithms.
- Quantum Variational Rewinding for Time Series Anomaly Detection. arXiv preprint arXiv:2210.16438 (2023).
- A quantum generative model for multi-dimensional time series using Hamiltonian learning. arXiv preprint arXiv:2204.06150 (2022).
- Live Demo of algorithm – https://pennylane.ai/qml/demos/tutorial_univariate_qvr.html.
Host: Walter Lambrecht