Teacher: Mirco Musolesi (DISI, University of Bologna) Teaching period: April 3rd, 20th, 27th 2021 and May 4th, 11th 2021 Total hours: 10 Assessment method: by project
In this module we will cover advanced topics in (Deep) Reinforcement Learning, such as function approximation methods with policy networks and actor-critic architectures, and Multi-Agent Reinforcement Learning. The module will cover both theoretical foundations and applications. We will discuss key recent papers in this area and we will outline the open challenges in this field.
Teacher: Paolo Torroni (DISI, University of Bologna) Teaching period: April 27th 2021 and May 4th, 11th, 18th, 25th 2021 Total hours: 10 Assessment method: by project
Argumentation studies how assertions are proposed, discussed, and resolved in the context of issues upon which several diverging opinions may be held. This is relevant to many disciplines, such as logic, philosophy, language and rhetoric, psychology, sociology, law, communication studies, an many more. In recent decades, computational argumentation has become a focus of interest in artificial intelligence, as it was recognized as a powerful paradigm for representing knowledge, operationalizing formal reasoning, framing multi-agent dialogues, and help conflict-resolution and decision-making. More recently, such an interest has expanded with the introduction of new applications of computational argumentation, and the increasingly important contribution of language technologies for the automated analysis and generation of natural arguments. This series of lectures aims to introduce the audience to this lively area, provide a broad overview of paradigms and methods, and illustrate applications in the context of explainable AI and argument mining.
Teacher: Michele Colajanni (DISI, University of Bologna) Teaching period: June 22nd, 24th, 29th 2021 and July 1st 2021 Total hours: 12 Assessment method: by project
A fully digital and interconnected world is changing forever the boundaries of any organization and citizen but is also favoring a proliferation of malicious cyber behaviors. The incessant dissemination of new products and services is taking too much priority over security and reliability mechanisms. A similar scenario requires a disruptive refoundation of cybersecurity solutions. The course focuses on the following topics: • 5G+IoT+Big Data = Unavoidable path towards cloudification • Rethinking the authentication mechanisms • The eternal problem of vulnerable software • The role of machine learning in cyber security
The course is open to PhD students in computer engineering and science, but other PhD students with scientific background are welcome. No specific competences on cyber security are required. All participants will be invited to re-think their research activities in terms of transparent cybersecurity.
Teacher: Roberto Casadei (DISI, University of Bologna) Teaching period: September - October2021 Total hours: 10 Assessment method: by project
The pervasiveness of computing and networking is creating significant opportunities for building valuable socio-technical systems. In particular, we are witnessing the emergence of artificial collectives, i.e., (large-scale) ensembles of (partially) autonomous interacting devices. Examples include computational ecosystems (like cloud data centers), crowds of augmented people, swarms of robots. Indeed, beyond what can be provided by individual smart devices, cyber-physical collectives can enable services or solve complex problems by leveraging a “system effect” emerging while coordinating and adapting to environment change. Therefore, understanding and building systems exhibiting collective intelligence and autonomic capabilities represents a prominent research goal, with contributions from several fields ranging from coordination to swarm intelligence, from multi-agent to self-* systems. In this short cycle of lectures, we will take a look at concepts and approaches contributing to the engineering of (computational) collective intelligence, or so-called collective adaptive systems. In the second part, we will focus on aggregate computing, a toolchain-backed macro-programming paradigm to model collective behaviour as a functional composition of field computations, and perform experiments in the ScaFi-Alchemist framework.
Teacher: Maria Pia Torricelli, Rosalia Miceli (Biblioteca di Ingegneria e Architettura, University of Bologna) Teaching period: February 1st, 8th, 9th, 15th, 16th 2021 Total hours: 21
Teacher: Daniela Loreti (DISI, University of Bologna) Teaching period: December 9th, 10th,15th, 16th 2020 Total hours: 10 Assessment method: by project
The course will provide an introduction to the main techniques to manage the complexity of High-Performance Computing infrastructures while taking advantage of their great computing power.
Teacher: Emanuele Rodolà (University of Rome, Sapienza) Teaching period: December 2nd, 4th, 8th, 11th 2020 Total hours: 10 Assessment method: by project
The course will cover the mathematical and computational foundations of geometric deep learning, going all the way from the basics to more recent advances in the vision, geometry, pattern recognition and learning communities. The final goal is to provide a broad understanding of this topic, and give the tools to analyze progress in this thriving area and possibly contribute to its advancement.