The aims of modern-day nuclear physics experiments are well summed up by the Jefferson Lab’s motto: ‘exploring the nature of matter’. To this end, the CLAS detector detects particles initially produced in collisions at the Jefferson Lab. This constitutes an event, and as thousands are produced by CLAS, only the relevant ones must be selected to investigate a given physical process. A technique called cut-based analysis is used to select relevant events. Conversely, machine learning is promising to revolutionise the field of data analysis. Could it outperform standard event selection techniques? To answer this question, this article will describe the implementation of a multilayer perceptron (MLP) for event selection at CLAS, by first presenting the fundamentals of particle physics, the CLAS detector, and machine learning. This article will then explain how a benchmark value was produced using cut-based analysis and compared with the output of machine learning event selection, demonstrating that machine learning algorithms can outperform the standard techniques used for event selection at CLAS. A better event selection process will improve the statistics of studies made using CLAS data, which will help answer some of the most fundamental questions of 21stcentury nuclear physics. For example, the discovery of exotic particles such as pentaquarks would allow a more detailed study of the strong force that binds nucleons, and might even shed light on the physics of neutron stars.