Off-the-shelf algorithms for time series classification and extrinsic regression

• Grant number: 25/04971-8
• Support Opportunities: Scholarships in Brazil - Master
• Start date: May 01, 2025
• End date: April 30, 2027
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: Diego Furtado Silva
• Grantee: Leonardo Rossi Luiz
• Host Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil
• Associated research grant: 22/03176-1 - Machine learning for time series obtained in mHealth applications, AP.PNGP.PI

Abstract

This research proposes developing and evaluating deep neural networks for analyzing time series in healthcare applications, focusing on data from real-world data repositories such as PhysioBank. Initially, experimental studies will be conducted to compare classification and extrinsic regression algorithms, identifying data standardization guidelines and assessing their advantages and limitations in the healthcare domain. As an alternative to the costly manual exploration of algorithms, we propose using meta-learning techniques and Bayesian optimization to automatically recommend models and hyperparameters based on data characteristics. The research will be conducted using data from specialized medical equipment, ensuring lower susceptibility to noise and external artifacts, and will later be extended to mHealth data.