Adversarially-trained autoencoders for robust unsupervised new physics searches

Blance, Andrew; Spannowsky, Michael; Waite, Philip

Abstract

Machine learning techniques in particle physics are most powerful when they are trained directly on data, to avoid sensitivity to theoretical uncertainties or an underlying bias on the expected signal. To be able to train on data in searches for new physics, anomaly detection methods are imperative, which can be realised by an autoencoder acting as an unsupervised classifier. The last source of uncertainties affecting the classifier are then experimental uncertainties in the reconstruction of the final-state objects. To mitigate their effect on the classifier and to allow for a realistic assessment of the method, we propose to combine the autoencoder with an adversarial neural network to remove its sensitivity to the smearing of the final-state objects. We quantify its effect and show that one can achieve a robust anomaly detection in resonance-induced tt¯ final states.

Citation

Blance, A., Spannowsky, M., & Waite, P. (2019). Adversarially-trained autoencoders for robust unsupervised new physics searches. Journal of High Energy Physics, 2019(10), Article 047. https://doi.org/10.1007/jhep10%282019%29047

• Journal Article Type: Article
• Acceptance Date: Sep 20, 2019
• Online Publication Date: Oct 4, 2019
• Publication Date: Jan 1, 2019
• Deposit Date: Oct 18, 2019
• Publicly Available Date: Oct 18, 2019
• Journal: Journal of High Energy Physics
• Print ISSN: 1126-6708
• Electronic ISSN: 1029-8479
• Publisher: Scuola Internazionale Superiore di Studi Avanzati (SISSA)
• Peer Reviewed: Peer Reviewed
• Volume: 2019
• Issue: 10
• Article Number: 047
• DOI: https://doi.org/10.1007/jhep10%282019%29047
• Public URL: https://durham-repository.worktribe.com/output/1317733