Mousam Rai is an Experimental Physics PhD student at University of Warwick under the supervision of Professor John Marshall. He has a background in Neutrino Physics, having achieved his BSc in Theoretical Physics at University of Leeds and MSc in Physics at University of Surrey. Currently a member of Deep Underground Neutrino Experiment (DUNE), he specialises in using machine learning techniques to improve characterisation performance of existing pattern recognition software. The need for greater knowledge on machine learning and its application is what motivated Mousam to be a Turing Doctoral Student.
Mousam is looking to improve the performance of Pandora Pattern Recognition Software which will be used at DUNE to characterise between "tracks" and "showers". These topologies appear when a fundamental particle called neutrino, interacts with the Argon atom inside the detector volume and the resulting particles transverse through the detector. Being able to correctly characterise and separate these so-called tracks and showers is very important as it will help physicists understand and maybe even answer fundamental problems such as matter-antimatter symmetry, neutrino mass hierarchy, and nucleon decay.
The problem of track and shower separation is a very difficult image processing problem which has been explored using Boosted Decision Trees which are typically used in experimental particle physics. By implementing deep learning techniques to tackle this problem, Mousam is looking forward to take a novel step towards tackling this problem.