Deep Learning: the new big trend in AI

Marco Lippi - Università di Bologna, 22 Aprile 2016

Abstract: The last decade has seen a huge progress in a wide variety of applications related to the area of artificial intelligence, from speech recognition to
computer vision, from natural language understanding to machine translation. Most of these improvements have been achieved by Deep Learning, a big trend in recent AI research, which goes back to multi-layered neural network architectures inspired by the human brain. Big companies such as Facebook, Google, Microsoft and Baidu are investing more and more resources in order to push forward with this novel technology. In this seminar, I will present some of the models and algorithms developed in the context of Deep Learning, together with some applications and a few criticisms which have recently been raised.


Slide del seminario zip


An Introduction to Complex Systems Science

Andrea Roli - Università di Bologna, 13 Maggio 2016

Complex system science (CSS) is a fairly recent field of science studying how parts of a system give rise to the collective behaviours of the system, and how the system interacts with its environment. It focuses on certain questions about parts, wholes and relationships. CSS is the corpus of theories and methods that can help dealing with complex systems (CS). Examples of CS are the brain, the society, the ecosystem, the cell, the ant colonies, the stock market. Typically, CS are have some of the following characteristics: nonlinear interactions, positive and negative feedbacks, adaptivity, evolvability, robustness, levels of organisation (i.e. tangled hierarchies). CSS has the objective of studying, modelling, and controlling CS.
In this talk, I will illustrate the principles of CSS emphasising its methods for studying and modelling system dynamics and dynamical regimes (order vs disorder), emergence and self-organisation. These concepts will be illustrated by means of prominent examples, such as the Ising model, Boolean networks and information diffusion models. Finally, the use of information theory in CSS will be presented, along with some recent results.


Slide del seminario pdf