WNN Predictors Prefetcher & Drones
Duration: 2026 -
Description: The project aims to evaluate how hardware implementations of Weightless Neural Networks can be leveraged to provide low inference latency in critical applications. Initially, the project intends to implement these neural network predictors for microarchitectures, specifically expanding our ongoing research on predictors and preprocessors. Another, more exploratory aspect, planned for the second year of the project, is the integration of these networks into drones using FPGAs to support autonomous flight command and control.
Subprojects:
Design, evaluation, and integration of weightless neural network models in prediction tasks on the RISC-V microarchitecture;
Design, evaluation, and implementation of weightless neural network models for online object and route detection, for use in autonomous drone control and command missions.
Pesquisadores: Diego Leonel Cadette Dutra (Coordinator, Professor); Priscila M. V. Lima (Professor); Lizy K. John (Professor); Felipe M. G. França (Senior Researcher); Andre Rotava (D.Sc. Student); Lucas B. Storino (M.Sc. Student); Aline Galdino de Oliveira (M.Sc. Student); Vivian Maria da Silva e Souza (Undergrad Student).
