Visualization in deep learning based image classification tasks

• Grant number: 20/07089-0
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: August 01, 2020
• End date: September 30, 2021
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Agreement: Microsoft Research
• Principal Investigator: Nina Sumiko Tomita Hirata
• Grantee: Jônatas de Souza Nascimento
• Host Institution: Instituto de Matemática e Estatística (IME). Universidade de São Paulo (USP). São Paulo , SP, Brazil
• Company: Universidade de São Paulo (USP). Instituto de Matemática e Estatística (IME)
• Associated research grant: 17/25835-9 - Understanding images and deep learning models, AP.PITE

Abstract

Deep neural networks are able to not only optimally predict the output but also to transform raw input data into more suitable representations. Such flexibility is accompanied with a large number of parameters that hinder the understanding of network functioning. To help understanding, visualization techniques are employed to gain insight or reinforce evidences that a network is working properly or not. In this project we aim to study and experiment with the diverse visualization based approaches and apply them as support for the development of deep learning based solutions for image classification tasks. (AU)