I was working on the algorithm of this article : Quantum detection of time series anomalies
Which is very interesting of course but I had one particular question : where is quantum computing supposed to be relevant in this model compared to classical methods for anomaly detection in time series ?
Hey @Delon! Welcome to the forum
This demo was a collaborative effort with external contributors, so I don’t have direct contact with them to get a better answer to your question. That said, I think the motivation for applying quantum to this problem is straightforward:
- time series anomaly detection is a well-known application in traditional machine learning, so trying a QML algorithm is a reasonable curiosity to have
- the particular algorithm used — QVR — is nice because it’s for gate model quantum computers (tied to hardware) and seems reasonably suited for time series data given its structure.
Hope this helps!
Thank you for your answer !
Yeah I think it is very interesting especially with NISQ computers ! I didn’t find any other algortihm for QML anomaly detection in time series.
But I think it would be interesting to point out where the quantum computer brings “an advantage” compared to classical methods.
Interesting! You have to be quite careful with “advantage”, though . I think this demo is mostly proof of concept — there’s no blatant claims of quantum advantage.