Professor Francesco Orabona will present the following seminar today from 2:30 in NCS, Room 120

Title: From Machine Learning to Online Learning

Machine learning is becoming nowadays an essential tool for any process that requires to model/predict/cluster huge amount of raw data.
As essential as they are, standard machine learning algorithms still require the data to obey to a number of (unreasonable) assumptions, and the need for a human to supervise the learning process.

In this talk, I will briefly introduce the standard machine learning protocol and its extension to the more generic and powerful Online Learning setting. As a sample of the potentialities of online learning, I will briefly present a learning algorithm that work under adversarial assumptions and self-tunes its parameters.

Prior knowledge of machine learning notions is not required.