The course will go into detail about modern Artificial Intelligence techniques used in research and production, focusing both on the theory behind the algorithms and on practical and concrete examples using the Marconi100 supercomputer of Cineca. The codes that will be written and executed during the course will concern both training and prediction using artificial neural networks on a large and small scale. 

Topics covered:

Data Analysis and Deep Learning
Basic HPC usage
HPC for AI: Distributed Training
Adaptation of pre-trained models
Natural Language Processing & Semantic Retrieval
Computer Vision and Generative Adversarial Network (GAN)
 

Prerequisites:

Mathematics and linear algebra skills
Knowledge of Python
Ability to run code on Linux machine
Basic knowledge of Machine Learning and Neural Networks
 

Output Skills:

Knowledge of machine learning and data management algorithm examples.
Knowledge of HPC benefits for ML and AI
Knowledge of training and hyperparameters 
Knowledge of the concept of fine-tuning and transfer-learning for adaptation to specific cases 
Knowledge of how to run AI on HPC
Theoretical and practical knowledge of distributed training
 

Lecturers:

Federico Betti (Researcher at LeonardoLabs)
Domiziana Catalano (Researcher at LeonardoLabs)
Eric Pascolo (HPC Specialist at Cineca)
Laura Morselli (HPC Specialist at Associazione Big Data)