Model Independent Searches for New Physics using Machine Learning
Our main idea is to assess the performance of different semi-supervised or unsupervised methods using different benchmark data sets. Besides commonly explored multi-jet signatures, we also want to consider multi-lepton processes. In order to compare the performance of different methods, we also consider several types of data input, e.g., kinematic variables and image data sets. This will provide a comparison of different methods on the same footing.
Community Needs, Tools and Resources
(Massachusetts Institute of Technology)