Full text
| |
| Author(s): |
Danilo Barbosa Coimbra
Total Authors: 1
|
| Document type: | Doctoral Thesis |
| Press: | São Carlos. |
| Institution: | Universidade de São Paulo (USP). Instituto de Ciências Matemáticas e de Computação (ICMC/SB) |
| Defense date: | 2016-06-17 |
| Examining board members: |
Fernando Vieira Paulovich;
Hugo Alexandre Dantas do Nascimento;
Maria Cristina Ferreira de Oliveira;
Paulo Aristarco Pagliosa;
Alexandru Cristian Telea;
Ricardo da Silva Torres
|
| Advisor: | Fernando Vieira Paulovich; Alexandru Cristian Telea |
| Abstract | |
The continuously advent of new technologies have made a rich and growing type of information sources available to analyses and investigation. In this context, multidimensional data analysis is considerably important when dealing with such large and complex datasets. Among the possibilities when analyzing such kind of data, applying visualization techniques can help the user find and understand patters, trends and establish new goals. Some applications examples of visualization of multidimensional data analysis goes from image classification, semantic word clouds, cluster analysis of document collection to exploration of multimedia content. This thesis presents several visualization methods to interactively explore multidimensional datasets aimed from specialized to casual users, by making use of both static and dynamic representations created by multidimensional projections. Firstly, we present a multidimen- sional projection technique which faithfully preserves distance and can handle any type of high-dimensional data, demonstrating applications scenarios in both multimedia and text docu- ments collections. Next, we address the task of interpreting projections in 2D, by calculating neighborhood errors. Hereafter, we present a set of interactive visualizations that aim to help users with these tasks by revealing the quality of a projection in 3D, applied in different high dimensional scenarios. In the final part, we address two different approaches to get insight into multimedia data, in special soccer sport videos. While the first make use of multidimensional projections, the second uses efficient visual metaphor to help non-specialist users in browsing and getting insights in soccer matches. (AU) | |
| FAPESP's process: | 11/17925-1 - Employing Visualization Techniques for Multimodal Data Analysis |
| Grantee: | Danilo Barbosa Coimbra |
| Support Opportunities: | Scholarships in Brazil - Doctorate |