Resource-constrained neural models for time series classification and extrinsic regression

• Grant number: 24/09747-6
• Support Opportunities: Regular Research Grants
• Start date: October 01, 2024
• End date: September 30, 2026
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Mobility Program: SPRINT - Projetos de pesquisa - Mobilidade
• Principal Investigator: Diego Furtado Silva
• Grantee: Diego Furtado Silva
• Principal researcher abroad: Gustavo Enrique de Almeida Prado Alves Batista
• Institution abroad: University of New South Wales (UNSW), Australia
• Host Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil
• Associated researchers: Ricardo Cerri ; Ricardo Marcondes Marcacini
• Associated research grant: 22/03176-1 - Machine learning for time series obtained in mHealth applications, AP.PNGP.PI

Abstract

Time series has become a fundamental data type in many applications, given the increasing pervasiveness of devices capable of collecting and storing temporal data. The most notable recent advances in machine learning for time series lie in the proposal of deep learning architectures. However, these neural networks usually rely on an enormous number of parameters, which makes them computationally costly. On the other hand, many practical applications depend on lightweight models due to hardware limitations. In this scenario, this research project proposes investigating different techniques for building resource-constrained neural models for extrinsic classification and regression of time series. In particular, health is this research's primary application domain of interest. (AU)