Statistical Learning Algorithms for Forensic Scientists

The goals of this workshop are to: (1) introduce attendees to the basics of supervised learning algorithms in the context of forensic applications, including firearms and footwear examination and trace evidence, while placing emphasis on classification trees, random forests, and, time permitting, neural networks; (2) introduce the concept of a similarity score to quantify the similarity between two items; (3) show how learning algorithms can be trained to classify objects into pre-determined classes; (4) discuss limitations of Machine Learning (ML) algorithms and introduce methods for assessing their performance; and (5) discuss the concept of a Score-based Likelihood Ratio (SLR): computation, advantages, and limitations.