Rocco De Rosa curriculum vitae EDUCATION 2011–2014 PhD - Mathematics and Statistics for Computational Science, Università degli studi di Milano, Milan, supervised by Nicolò Cesa-Bianchi. Thesis: Confidence-based and Nonparametric Classification: Applications to Credal Classification, Online and Active Learning. 2009–2011 MSc (110/110 cum laude) - Informatics, Università degli studi di Milano, Milan. Thesis: “Time series classification by precise and imprecise Markov models with applications in action recognition” 2001–2003 BSc - Digital Communications, Università degli studi di Milano, Milan. 1996–2001 Upper School - Informatics, I.T.I.S Stanislao Cannizzaro, Milan. RELEVANT WORK EXPERIENCE 2015–current Data Scientist - Business Intelligence, KeyOS - Sisal , Milan. Machine Learning solution for: Maximizing the Player Lifetime Value. Skill acquired: { Predictions with Partial feedback (multi-armed bandit algorithms); { Recommendation Systems; { Personalized Promotions; 2014–current Postdoctoral Researcher, Università degli studi di Milano, Milan, supervised by Nicolò Cesa-Bianchi. Online Learning - Time Series Mining - Real Time Bidding. Skill acquired: { Predictions with Partial Feedback (bandit algorithms); { Recommendation Systems; { Scalable prediction algorithms (Stream Data Mining); { Real Time Bidding;. 2014–2014 Research Assistant, Oxford Brookes Vision Group, Oxford. Computer Vision. Gesture-Action recognition. Skill acquired: { Feature extraction from video frames; { Real Time Activity Recognition; { Multilinear classification (linear classifier where each sample has more than one label). 2012–2013 Data Scientist, Agenzia delle dogane, Rome. Risk analysis of customs bills. Skill acquired: { Analysis of complex system; { Data integration and transformation from heterogeneous source data; { Real-time automated detection of trade customs frauds based on the analysis of customs declarations. 2011–2012 Research Assistant, Swiss RE Ltd, Zurich. Text mining applications and complex data extraction. Skill acquired: { Auto Categorization of documents. { Data Extraction from documents. 2010–2013 Research Assistant, IDSIA, Lugano. Time series classification with imprecise models. Skill acquired: { Robust classification algorithms; { General imprecise probabilistic methods (credal sets, convex optimizations, utility measures). OTHER WORK EXPERIENCE 2013–2013 Informatics consulting, Esalpi, Milan. Optimized Web application for smartphone. 2009–2010 Informatics consulting, Links, Milan. Java Desktop application. 2007–2008 Software designer, Linkas, Milan. Java J2ME mobile application. 2004–2007 Analyst-Developer, Esalpi, Milan. Java Web enterprise application. 2003–2004 Developer, Accenture, Milan. Java Web and Desktop application. 2002–2003 Junior developer, Plug-in, Milan. Visual basic Desktop application. LANGUAGES Language mother tongue Italian Language intermediate English KEY SKILLS IT skills { Experienced in programming in Java, C++, MATLAB, and using database such as Sql server , MySQL and Access; { Proficient in many Microsoft Office packages, UNIX and Windows operating systems; { Experienced in Data Mining Frameworks: WEKA (Data Mining Software in Java), MOA (MASSIVE ONLINE ANALYSIS), Vowpal Wabbit (Scalable implementation of Online Machine Learning). Data Mining { KEYwords (Field|Application): Data Mining (Data Analysis), Classification (Pheskills nomena Predictions), Clustering (e.g. Customer Segmentation), Recommendation System (e.g. Rating - Preferences User-Products), Filtering (e.g. Spam Filtering, Real-time automated detection of frauds),Real Time Bidding (e.g. place Automatically Auction Reserve Price), Computer Vision (e.g. Human Activity Recognition on Video Stream, Continuous Activity Recognition), Text Mining (Auto Categorization of Documents), Big Data (Scalable Prediction Models), Learning with Partial Feedback (e.g. Optimizing Online Display Ads, Ad Click Prediction), Active Learning (Save expensive annotation process in prediction models learning, rare class detection (fraud detection)). { Designer/analyst/programmer for Data Mining problems: data analysis, data extraction and transformation, feature engineering, model selection, model tuning, experiment setting, testing protocol, report and results analysis; { Know-how in statistical learning theory: major classification algorithms (Bayesian methods, general linear models, decision trees, time series classification, credal classification), classification paradigms (batch, online, incremental, active, selective sampling, filtering, predictions with Partial feedback), optimization tasks (convex, quadratic, heuristics), major clustering techniques (partitioning methods, hierarchical methods, density-based methods), probabilistic models (Bayesian models, Imprecise probabilities models, Markov models, frequentist models). People & { Supervised the laboratory project of an undergraduate, both day-to-day running Project of the project and its overall planning; Management { Mentor to other postgraduate students. Problem { Working across different scientific areas of Machine Learning, demonstrated solving independent though in analysing problems, adopting suitable strategies and developing new techniques. The results are being reviewed for publication (see my website). INTEREST AND ACTIVITIES Sports runner, skier, walk in nature Chess amateur player Travel an avid backpacker traveller