In this project, I investigated the feasibility of using Convolutional Neural Networks to detect and identify supernova neutrinos events in liquid argon time-projection chamber (a new type of detector).
Implemented a simulation for electronic noise in detectors and developed a machine learning classifier to discern 'clean' simulated neutrinos from noise-augmented slices. The project further delves into the analysis of the classifier's performance across varying noise intensities.
Noise Simulation: Created a method to simulate the electronic noise expected in the detector, which follows a normal distribution.
ML Classifier Development: Developed a classifier using Convolutional Neural Networks (CNN) to segregate 'clean' simulated neutrinos from 'empty' slices inundated with 'electronic noise'.
Performance Analysis: Assessed the machine learning algorithm's robustness by evaluating its performance across different noise levels.
Classifier Testing: The classifier was tested on simulated neutrinos overlaid with varying levels of noise to understand the threshold of noise that impacts its accuracy.