ML Algorithm - IceCube
Developed machine learning algorithm to predict energy and orientation of high-energy particles detected by the IceCube Neutrino Observatory.
Skills
Machine Learning, Python, TensorFlow, Physics, Data Analysis, Particle Physics, Scientific Computing
During my internship at the Technical University of Munich, I worked with data from the IceCube Neutrino Observatory, a massive particle detector embedded in the Antarctic ice at the South Pole.
IceCube consists of a cubic kilometer array of photomultiplier detectors arranged in a three-dimensional grid throughout the ice. When high-energy particles, particularly neutrinos, interact with the ice, they produce secondary particles that emit light as they travel through the detector array. These light patterns create distinctive "trails" that can be measured by the detectors.
My contribution to this project involved developing a 3D machine learning algorithm that could process the detection data and predict two critical properties of the original particle: 1. The directional origin of the particle 2. The energy level of the particle
This project represented my first significant exposure to applied machine learning in a scientific research context. Working with actual data from one of the world's most unique astronomical instruments provided invaluable experience in data processing, model development, and the practical application of machine learning to solve real-world physics problems.