Efficient Code Generation for Deep Learning Models in Scientific Computing

• Grant number: 25/00406-4
• Support Opportunities: Scholarships in Brazil - Doctorate
• Start date: April 01, 2025
• End date: March 31, 2026
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computer Systems
• Principal Investigator: Guido Costa Souza de Araújo
• Grantee: Thais Aparecida Silva Camacho
• Host Institution: Instituto de Computação (IC). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
• Associated research grant: 13/08293-7 - CCES - Center for Computational Engineering and Sciences, AP.CEPID

Abstract

Deep learning models have become crucial in various domains and applications that benefit from scientific computing. Furthermore, the execution time of these models is critical to meeting the ever-growing demands in many areas. One example is the problem of predicting molecular properties, where the model's execution time is crucial when the system needs to perform multiple real-time predictions. In this context, deep learning compilers enable automatic high-performance code generation for various deep learning models on different devices. However, existing compilers have certain limitations that impede the optimization of models. The main goal of this Ph.D. project is to improve the usability, code quality, and compilation time of the deep learning compilers.