School of
Information Technology and Electrical Engineering

Granular Computing Methods for Learning and Mining on Structured and Unconventional Data

Lorenzo Livi, SAPIENZA - University of RomeMon, 18/06/2012 - 14:00
Prof. Xiaofang Zhou

Similarity (and dissimilarity) measures are ubiquitous in different intelligent data processing fields. Application of such measures constitutes the key element to construct data driven modeling systems that can handle a broad range of input domains. Indeed, many interesting practical applica- tions deal directly with non metrical patterns, such as images, audio/video signals, and bio-chemical compounds. In such cases, where the metrical structure of the input space D is not obvious, the role of similarity measures is to provide a way to numerically quantify the commonalities among the elements of D.
Granular Computing (GrC) is an emerging and unifying framework that incorporates many data abstraction and representation methods. One of the main objectives of a granular computing system is to represent compactly the elements of D that are in some sense indistinguishable at the current level of abstraction. Those groups of low level entities are called information granules. However, designing accurate GrC systems is costly from the computational viewpoint.
In this talk, I will discuss how to combine (dis)similarity measures and GrC modeling techniques to conceive optimized inductive inference engines able to efficiently deal with different input domains. I will focus on patterns defined by labeled graphs, sequences, and fuzzy sets.


Lorenzo Livi received his bachelor (2007) and master (2010) degree both from the Computer Science department of SAPIENZA University of Rome. During his studies, he has worked as ICT analyst in different companies in Rome, mainly in the field of database design. Currently, he is a PhD student in the Information and Communication Science doctoral program, at the department of Information Engi- neering, Electronics and Telecommunications of the same University. His supervisor is Prof. Antonello Rizzi. He is currently spending 7 months as visiting PhD student at the Computer Science department of Ryerson University, Toronto, under the supervision of Prof. Alireza Sadeghian. His main research inter- ests are focused on Pattern Recognition, Soft Computing, and Parallel Computing, mostly considering graph-based problems.

Seminar Type: 

ITEE Research Seminar