Seminario pubblico di Aristides Anagnostopoulos
Speaker:
Aristides Anagnostopoulos
speaker DIAG:
Data dell'evento:
Thursday, 24 February, 2022 - 17:00
Luogo:
DIAG - Aula Magna
Contatto:
Stefano Leonardi
in ottemperanza ai requisiti previsti dalla procedura selettiva per n. 1 posto di Professore di ruolo di I Fascia per il Settore Concorsuale 09/H1 – Settore Scientifico Disciplinare ING-INF/05 presso il Dipartimento di Ingegneria Informatica, Automatica e Gestionale “Antonio Ruberti”, codice concorso 2021POR039 del 14/07/2021,
Giovedì 24 febbraio 2022
ore 17:00, Aula Magna, DIAG
e in collegamento Zoom
Giovedì 24 febbraio 2022
ore 17:00, Aula Magna, DIAG
e in collegamento Zoom
ID: 89988012270
Passcode: 133564
Passcode: 133564
si terrà il seminario pubblico di Aristides Anagnostopoulos sulle sue attività di ricerca. La notizia è pubblicata anche sul sito del DIAG.
===
Title: Data-Science Applications in Biostatistics and Philology
Anstract:
The current availability of data from various sources together with the development of new AI methods for analyzing them, has started transforming multiple disciplines in the physical and life sciences, and even in the humanities. Of particular importance, is the recent development of explainability approaches in AI, which make possible such multidisciplinary research, as they allow to identify the features in the data that are responsible for the results and present them to the experts in the field, who in turn can interpret them and guide the design choices of the system.
In this talk I will present some recent work that offers new ways of analysis in the areas of biostatistics and in philology. I will present a new method for genome-wide association studies (GWAS), which leverages the power of neural networks by finding epistatic interactions: combinations of genes that acting together may be responsible for a given trait disease, in our case hypertension. This method detects subsets of input features to a neural network that interact nonlinearly and have an effect to the output of the network. Next I will present a methodology for authorship attribution of texts written in medieval Italian. I will finish the main part of the talk with an approach for rank aggregation for selecting important features.
Finally, I will briefly discuss some other areas of my current research.
Bio: Aristidis (Aris) Anagnostopoulos (http://aris.me) is an Associate Professor at the Dept. of Computer, Control, and Management Engineering (DIAG) of Sapienza University of Rome. He graduated from the Computer Engineering and Informatics Department of Univesity of Patras, Greece, he obtained his Ph.D. from Brown University, and afterwards he did a postdoc at Yahoo! Research in Santa Clara, CA. Since 2008 he is at DIAG, as a Marie-Curie Fellow, and then as an assistant and associate professor. His main research interests include randomized and approximation algorithms as well as the application of machine-learning approaches on data-science problems.