Investigation of graph-based contextual measures for weakly-supervised learning

• Grant number: 20/08854-2
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: September 01, 2020
• End date: August 31, 2021
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computer Systems
• Agreement: Microsoft Research
• Principal Investigator: Daniel Carlos Guimarães Pedronette
• Grantee: Bionda Rozin
• Host Institution: Instituto de Geociências e Ciências Exatas (IGCE). Universidade Estadual Paulista (UNESP). Campus de Rio Claro. Rio Claro , SP, Brazil
• Company: Universidade Estadual Paulista (UNESP). Campus de Rio Claro. Instituto de Geociências e Ciências Exatas (IGCE)
• Associated research grant: 17/25908-6 - Weakly supervised learning for compressed video analysis on retrieval and classification tasks for visual alert, AP.PITE

Abstract

While traditional distance/similarity measures are mainly based on pairwise analysis, contextual measures also consider neighborhood similarity relationships. Recently, such measures have been successfully exploited in various unsupervised learning tasks, especially using graph-based approaches. In this scenario, this project considers the hypothesis that these measures can also be applied in weakly supervised learning tasks. The main idea consists in expanding small training sets through reliable similarity relationships identified in graphs. In this way, the main objective is to investigate if such approaches can achieve accuracy gains when applied on weakly supervised learning methods.

Scientific publications

(References retrieved automatically from Web of Science and SciELO through information on FAPESP grants and their corresponding numbers as mentioned in the publications by the authors)

ROZIN, BIONDA; PEREIRA-FERRERO, VANESSA HELENA; LOPES, LEONARDO TADEU; PEDRONETTE, DANIEL CARLOS GUIMARAES. A rank-based framework through manifold learning for improved clustering tasks. INFORMATION SCIENCES, v. 580, p. 202-220, NOV 2021. (17/25908-6, 20/02183-9, 20/08854-2, 18/15597-6)