ADFilter -- A Web Tool for New Physics Searches With Autoencoder-Based Anomaly Detection Using Deep Unsupervised Neural Networks

TL;DR

ADFilter is a web-based tool utilizing autoencoder neural networks for anomaly detection in collision data, aiding new physics searches by identifying events that deviate from the Standard Model.

Contribution

The paper introduces ADFilter, a novel web tool that applies deep unsupervised autoencoders to analyze collision events for anomaly detection, enhancing reinterpretation of LHC results.

Findings

Effective anomaly detection in collision data

Improved exclusion limits for new physics

Versatile application to LHC data analysis

Abstract

A web-based tool called ADFilter was developed to process collision events using autoencoders based on a deep unsupervised neural network. The autoencoders are trained on a small fraction of either collision data or Standard Model Monte Carlo simulations. The tool calculates loss distributions for input events, helping to determine the degree to which the events can be considered anomalous. It also calculates two-body invariant masses both before and after the autoencoders, as well as cross sections. Real-life examples are provided to demonstrate how the tool can be used to reinterpret existing LHC results with the goal of significantly improving exclusion limits.