Investigação de Abordagens Auto-Supervisionadas utilizando Medidas baseda em Ranqueamento

• Grant number: 23/08150-3
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: July 01, 2023
• End date: June 30, 2024
• Field of knowledge: Physical Sciences and Mathematics - Computer Science
• Principal Investigator: Daniel Carlos Guimarães Pedronette
• Grantee: Guilherme Henrique Jardim
• Host Institution: Instituto de Geociências e Ciências Exatas (IGCE). Universidade Estadual Paulista (UNESP). Campus de Rio Claro. Rio Claro , SP, Brazil
• Associated research grant: 18/15597-6 - Aplication and investigation of unsupervised learning methods in retrieval and classification tasks, AP.JP2

Abstract

Despite the significant progress made in supervised learning, the common requirement for extensive labeled datasets represents a severe bottleneck. In this scenario, other learning paradigms capable of addressing the challenge associated with the scarcity of labeled data represent a relevant alternative solution. Self-Supervised Learning (SSL) approaches have been established as a promising way to exploit machine learning advances through many unlabeled examples and without any human-annotated labels. A relevant advantage of SSL algorithms is their ability to leverage large amounts of unlabeled data without the need for human annotations to generate pseudo labels during training. Therefore, how to obtain effective pseudo-labels plays an important role and poses a key challenge for SSL approaches. On the other hand, a promising alternative to better represent the structure of datasets relies on the use of contextual-sensitive similarity measures, which have been successfully applied to capture the geometry of the underlying manifold. In this undergraduate research project, we intend to investigate the use of rank-based contextual similarity measures for self-supervised learning approaches.