Controlling the effects of user manipulation on multidimensional projections

• Grant number: 15/19851-6
• Support Opportunities: Scholarships abroad - Research Internship - Master's degree
• Start date: November 01, 2015
• End date: January 31, 2016
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: Fernando Vieira Paulovich
• Grantee: Samuel Gomes Fadel
• Supervisor: Alexandru-Cristian Telea
• Host Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil
• Institution abroad: University of Groningen, Netherlands
• Associated to the scholarship: 14/13010-7 - Applying machine learning to information visualization, BP.MS

Abstract

The large amount of available data on a diverse range of human activities provides many opportunities for understanding, improving and revealing unkown patterns about them.Powerful automatic methods for extracting this knowledge from data are already available from machine learning and data mining.They, however, rely on the expertise of analystis in order to improve their results when those are not satisfactory.Visual analytics alleviates this requirement by coordinating the use of automatic data mining and machine learning methods with data visualization to guide their refinement and tuning by providing additional insights to the analyst.In this context, interactive multidimensional projections are an effective tool for analyzing multidimensional data, revealing their underlying structure while allowing the analyst to manipulate the results to provide even further insight.This manipulation, however, has received little attention regarding their influence on the mappings.Since they incur a distortion which might not represent the original data faithfully, it is important to determine this distortion in order to improve their effectiveness.This is the main motivation for this research.We discuss the advantages of interactive multidimensional projections for analysis and highlight their current limitations, proposing a methodology to address these limitations based on the formulation of the distortion while employing visual cues to represent the distortion.