professore associato
Settore scientifico disciplinare: 
U14, Piano: P02, Stanza: 2046
Viale Sarca, 336 - 20126 MILANO
Orario di ricevimento: 

On Dating.


Fabio Stella is Associate Professor in the Department of Informatics, Systems and Communication at the University of Milano-Bicocca. He received his Ph.D. in Computational Mathematics and Operations Research at University of Milano in 1995. He worked as assistant professor in University of Milano from 1993 to 1998 before joining the University of Milano-Bicocca in 1998. He worked as research assistant in the EEC IMPROD project devoted to improve quality of semiconductors using the Data Mining methodology. From 1998 to 2000 he worked as research leader in the Pirelli P-Vision project to improve tyre performance using Data Mining. From 2000 to 2001 he was consultant for the Risk Management of Banca Intesa. His main research activities include; Bayesian networks, continuous time Bayesian networks, Feedforward neural networks, on-line portfolio selection algorithms and probabilistic topic models. He serves as referee for many International Journals; Journal of Approximate Reasoning, Expert Systems with Applications, BioData Mining, Journal of Biomedical Informatics, Pattern Recognition Letters, Data Mining and Knowledge Discovery, Information Processing and Management, European Journal of Operations research, and Quantitative Finance. From 2015, he is member of the Programme Committee of the Uncertainty in Artificial Intelligence conference. In 2017 he became member of the Programme Committee of the RecSys Conference and of the PAKDD Conference, while in 2018 he started to serve as reviewer for the NIPS Conference. From 2006, Fabio Stella teaches Probability and Statistics for Bachelor of Science in Informatics and from 2007 he teaches Data and Text Mining for Master of Science in Informatics at the University of Milano-Bicocca. He also teaches Data and Text Mining in foreign Universities. In 2016 he developed the Massive Online Open Course pathway titled Introduction to Data Mining. The pathway is freely available on the EduOpen platform.


  • Artificial Intelligence: main contributions concern probabilistic graphical models with specific reference to Bayesian networks, continuous time Bayesian networks and continuous time Bayesian classifiers. Theoretical contributions have been given for the structural learning problem of continuous time Bayesian networks under stationary and non-stationary frameworks. Such models have been applied to biology, immunology, medicine with specific reference to acute myocardial infarction, and post stroke rehabilitation as well as to the financial sector.
  • Machine Learning, Data and Text Mining: the main contributions concern topic models with specific reference to Latent Dirichlet Allocation for automatically extracting knowledge from natural language text. This class of probabilistic models has been the subject of theoretical developments to validate the quality of the extracted knowledge. These models have been applied to Recommendation Systems to improve recommendation explanation. Data and text mining have been used to address and solve the following problems; automatic labeling of documents, classification of cells from microarray expressions data.
  • Artificial Neural Networks: numerical properties of the learning problem on feed-forward neural networks has been addressed and studied. An algorithm to select the optimal structure of feed-forward neural networks has been designed and developed. This class of computational models has been applied to address and solve the following problems: computational finance, traffic flow forecasting, functional analysis of molecular dynamics, identification and signaling of adverse drug reactions.
  • On-line Optimization Algorithms: three new on-line computational algorithms for portfolio optimization have been designed and developed. These algorithms work under different investment settings, namely with and without transaction costs. Their, theoretical performance has been studied, analyzed and compared to that of state of the art algorithms.


  • Acerbi, E., Viganò, E., Poidinger, M., Mortellaro, A., Zelante, T., & Stella, F. (2016). Continuous time Bayesian networks identify Prdm1 as a negative regulator of TH17 cell differentiation in humans. SCIENTIFIC REPORTS, 6, 23128. Dettaglio
  • Stella, F., & Villa, S. (2016). Learning Continuous Time Bayesian Networks in Non-stationary Domains. THE JOURNAL OF ARTIFICIAL INTELLIGENCE RESEARCH. Dettaglio
  • Rossetti, M., Stella, F., & Zanker, M. (2013). Towards Explaining Latent Factors with Topic Models in Collaborative Recommender Systems. In 24th International Workshop on Database and Expert Systems Applications (pp.162-167). Dettaglio
  • Stella, F.A., & Amer, Y. (2012). Continuous Time Bayesian Network Classifiers. JOURNAL OF BIOMEDICAL INFORMATICS. Dettaglio
  • Gaivoronski, A., & Stella, F.A. (2003). On-line portfolio selection using stochasticprogramming. JOURNAL OF ECONOMIC DYNAMICS & CONTROL, 27, 1013-1043. Dettaglio