Andrea Asperti is Full Professor of Machine Learning at the Department of Computer Science and Ingeneering (DISI) of the University of Bologna. He has been Director of the Department and Coordinator of the university degree in Computer Science. For several years, he has been a member of the Advisory Committee of the World Wide Web Consortium. His researches have been focused on theory, programming languages, knowledge management and machine intelligence. He is Editor in Chief of the Journal of Formalized Reasoning and coordinated the development team of the interactive therem prover "Matita". He is author of a number of peer reviewed scientific articles, and coordinated several national and international projects.
Neural networks are a class of machine learning algorithms, originally inspired by the brain, structured in layers of interconnected artificial neurons.
The network can be trained on data to optimize its connections to a specific task.
Deep neural networks, that is networks with multiple internal (so called hidden) layers, have recently seen a lot of success at practical applications. They’re at the heart of production systems at companies like Google and Facebook for image classification, speech recognition, natural language processing, language understanding or robotics.
The course gives an overview of the foundational ideas and the recent advances in neural nets, explaining the potentialities of the topic for practical purposes. We shall cover supervised and unsupervised techniques, methods for visualizing and understanding the behavior on neural nets, as well as adversarial techniques to fool them. We shall also hints to recent applications in the field of reinforcement learning, and some amazing results in game simulation.