wnnParIO

Período: 2023 - 

Descrição: This project investigates the impact of weightless neural networks on the use of graph learning and new parallel I/O mechanisms for utility sector applications such as SDDP. In the context of graph learning, this project aims to propose and evaluate new architectures for comparison with state-of-the-art methods (deep neural networks, graph neural networks, and kernel methods). The impacts of improvements in graph learning on society are promising, enabling better and faster medical diagnoses, the development of new drugs and treatments, improved content recommendation, the resolution of non-polynomial logical and mathematical problems, and better machine learning in general. In the context of improvements in parallel applications such as SDDP (Stochastic Dual Dynamic Programming), the challenge is the efficient scalability of I/O in massive data parallel applications, especially in high-performance computing (HPC) environments.

Pesquisadores: Diego Leonel Cadette Dutra (Coordenador, Professor).