The course provides an introduction to artificial intelligence techniques for experimental and applied physics, such as machine learning and statistical learning. At the end of the course, the student will be able to create and train ad-hoc neural networks with KERAS and scikit-learn.
Course Prerequisites
Basic knowledge of object-oriented languages, as provided in the course "Metodi informatici della Fisica". Basic knowledge of radiation-matter interaction.
Teaching Methods
Oral lectures, assisted by presentations, that will be made available to students in pdf format via the Kiro platform.
Assessment Methods
Oral examination. Each student should provide a personal project in a selected field of interest, that entails the concepts introduced during the lectures
Texts
a) James at al., Introduction to Statistical Learning b) F. Chollet, Deep Learning with Python (ISBN: 9781617294433)
Contents
The course is divided into 2 modules: - Introduction to Deep Learning (I.Postuma) Brief introduction to Python and some essential packages to use KERAS. Introduction and development of neural networks to solve computer vision problems. - Machine learning in high energy physics (M. Pelliccioni) Data processing view of a large scale physics experiment. Detailed examples of ML applications in high energy physics experiments, with hands-on sessions based on scikit-learn and on the standard Python libraries, such as ML approaches to particle recognition in a detector, jet flavour classification or electron/hadron separation using decision trees.
