Keynote speech II: Supervised classification of noisy data
Modern digital technologies are now routinely used in many real world problems: from engineering applications to scientific investigations, digital equipments produce data that are very often incomplete or noisy. One of the most successful techniques in data analysis is supervised classification. It is based on the idea that, if data is known to belong to a certain number of classes, then it is possible to build a model capable to predict the class labels of incoming instances. We will describe techniques in the realm of fuzzy and robust classification algorithm that are resilient to noise present in data, providing some examples of their successful application.
Speaker: Prof. Mario Rosario Guarracino, Italian National Research Council, Italy
High Performance Computing and Networking Institute
National Research Council of Italy
CNR Research Area Via Pietro Castellino, 111 80131IT Naples, ITALY
Principal Investigator at Laboratory for Genomics, Transcriptomics and Proteomics Affiliated Faculty at National Institute for High Mathematics "F. Saveri" of Italy, and Center for Applied Optimization, University of Florida
Prof. Mario Rosario Guarracino is a staff researcher at High Performance Computing and Networking Institute (ICAR) of National Council for Research (CNR), coordinating a group of six researchers. He initially focused on distributed software infrastructure for biological and medical informatics applications. Since 2008, he is Principal Investigator at the Laboratory for Genomics, Transcriptomics and Proteomics (http://www--‐ labgtp.na.icar.cnr.it/), a public--‐private initiative for the molecular diagnosis and gene therapy of rare genetic diseases (theranostics). At the Laboratory, he leads a research group composed of six post docs, five graduate research fellows, four fellows in training, two students, three software developers, and one technician. In the next months I am going to hire five postdoc research fellows. He is also affiliated with Center for Applied Optimization at University of Florida since 2005 and Institute for High Mathematics F. Saveri since 2009.
Prof. Mario Rosario Guarracino’s personal research interests are data mining and statistical machine learning and its application to life sciences. His aim is to study and develop novel algorithms, methods and software tools based on mathematical programming and statistical learning theory that can handle the novel paradigm introduced by data science. The objective is to find novel analysis techniques that (i) are robust to errors and missing values, (ii) can handle big data varying over time and (iv) can use prior knowledge. He applies his expertise to biological problems in which data is produced by high throughput technologies, such as Microarray, Next--‐Generation Sequencers (NGS), Nuclear Magnetic Resonance (MNR) and Raman spectrometers. The results of his research are summarized in more than sixty peer--‐reviewed papers. He recently passed the qualifying exam as associate professor in statistics in Italy.