Predicting Jet Count in Proton-Proton Collisions using Machine Learning and Deep Learning Models

In the field of high-energy physics, analyzing and elucidating data from particle collisions at the CERN Large Hadron Collider (LHC) poses significant challenges due t o the complexity and high dimensionality of the data. Proton-proton collisions at the LHC primarily involve strong interactions in which quark degrees of freedom are involved, with electromagnetic processes characterized by small energy loss of incident protons resulting in small multiplicity of the final state. A correlation can be established between the impact parameter of the collision and the probability of strong or electromagnetic interaction. In this study, we apply various machine learning and deep learning models to predict the number of jets produced in these collisions, a key observable in many physics analyses at the LHC. Our models, which include K-Nearest Neighbors (KNN), Weighted K-Nearest Neighbors (W-KNN), Random Forest Classifier (RFC), Multi Layer Perceptron (MLP), and a custom Deep Learning model (DL), are trained and evaluated on the “CERN Proton Collision Prediction” dataset from Kaggle. Our results demon-strate the effectiveness of these models, with accuracies ranging from 81.00% to 93.92%. The RFC model outperformed all others, demonstrating the effectiveness of ensemble methods in handling high-dimensional data and reducing the risk of overfitting. These results underscore the potential of machine learning and deep learning techniques in the analysis and interpretation of complex data from particle collisions, and their ability to enhance our understanding of particle physics. Future work will focus on further optimization of the models, exploration of other models and feature engineering techniques, and application of these techniques to other problems in particle physics and beyond.