Creating and sharing knowledge for telecommunications

Project: Statistical and Structural Learning: Applications in Multimedia Data Mining and Retrieval

Acronym: SSL-AMDMR
Main Objective:
The focus are statistical and structural learning techniques. Exploratory analysis (mining) and content-based retrieval of multimedia data serve both as a practically relevant application domain, and as a source of challenging problems. The Goals to be pursued:
- new unsupervised learning criteria (i.e., simultaneous feature and model selection);
- unsupervised learning methods under a model-based framework and/or new clustering criteria;
- new dissimilarity measures, adequate for various types of representation (strings, quantitative, and hybrid), typically found in multimedia;
- combination of different types of representation in a coherent framework for similarity computation;
comparison and blending of different clustering techniques / criteria into a workable methodology;
- efficiency and scalability of learning algorithms.
Reference: POSI/33143/SRI/2000
Funding: FCT/POSI
Start Date: 01-01-2001
End Date: 31-12-2004
Team: Mario Alexandre Teles de Figueiredo, Ana Luisa Nobre Fred
Groups: Pattern and Image Analysis – Lx
Local Coordinator: Mario Alexandre Teles de Figueiredo
Associated Publications