Title: Event-based anomaly detection for new physics searches at the LHC using machine learning

Abstract: This paper discusses model-agnostic searches for new physics at the Large Hadron Collider (LHC) using anomaly-detection techniques for the identification of event signatures that deviate from the Standard Model (SM). We investigate anomaly detection in the context of machine-learning approaches using autoencoders, and illustrate expected shapes of invariant masses in the outlier region using Monte Carlo simulations. Challenges and conceptual limitations of this approach are discussed.