&nbsp;

Regular meetings and seminars will be an important part of the scientific program. Attendance to seminars can give rise to credits towards the completion of the taught part of the PhD program.

Each year, a number of seminars will be devoted to presentations of open problems suggested by companies or groups sponsoring the program. Scientists involved in the PhD program and invited guests will also give seminars on mathematical, statistical, modeling or legal tools that are deemed useful for the solution of such problems and for the general background of a data scientist.

March 17, 5:00PM, Room: Aula Magna (Dept. Mathematics) Ernesto De Vivo (Univ. Genova)
Machine Learning from a mathematical view point: a gentle introduction.

Abstract: The talk is devoted to a mathematical introduction to Machine Learning. We focus on supervised learning setting and kernel methods with square loss by using classical tools from the theory of inverse problems.

Thanks to CRT Foundation, Project "Strengthening and internationalization of advanced training in Data Science at the University of Turin"

March 24, 5:00PM, Room: Aula Magna (Dept. Mathematics), Giacomo Aletti (Univ. Milano)
Mathematical, numerical and statistical problems in the analysis of probabilistic counting of distinct elements

Abstract:Data streams are sequences of objects that cannot be available for random access but must be analyzed sequentially when they arrive and immediately discharged. One of the main applications in streaming algorithms concerns the problem of counting the number F0 of distinct elements in a stream, when the information that can be stored is of the order of the logarithm of the quantity of interest F0. In this seminar, the Flajolet-Martin class of probabilistic algorithms is analyzed from a probabilistic point of view, thus opening several computational, statistical, and mathematical problems. Connections with some consolidated theories will be shown. The results presented may be found in Aletti G. Analytical Confidence Intervals for the Number of Different Objects in Data Streams. Big Data Research, Volume 25, 2021

Thanks to CRT Foundation, Project "Strengthening and internationalization of advanced training in Data Science at the University of Turin"

2022