Joaquim Gomes Da Costa Eulalio De Souza

Title: Joaquim Gomes Da Costa Eulalio De Souza: Innovator in Neural Network Processing

Introduction

Joaquim Gomes Da Costa Eulalio De Souza is a notable inventor based in Porto Alegre, Brazil. He has made significant contributions to the field of neural network processing, particularly through his innovative patent that enhances traditional chip processing methods. His work is pivotal in advancing the efficiency of neural network accelerators.

Latest Patents

Joaquim holds a patent titled "System and method for processing convolutions on crossbar-based neural network accelerators for increased inference throughput." This patent outlines systems and methods designed to improve traditional chip processing. By utilizing crossbar computations, the convolution layer can be flattened into vectors, which are then organized into a matrix. Each submatrix of the input corresponding to a convolution window is also flattened into a vector. The convolution is computed as the dot product of each input vector and the filter matrix. The method innovatively uses intra-crossbar and inter-crossbar computations to optimize the processing of convolution iterations.

Career Highlights

Joaquim is currently employed at Hewlett Packard Enterprise Development LP, where he continues to work on cutting-edge technologies. His expertise in neural networks and chip processing has positioned him as a valuable asset in the tech industry.

Collaborations

Throughout his career, Joaquim has collaborated with esteemed colleagues, including Glaucimar Da Silva Aguiar and Francisco Plínio Oliveira Silveira. These collaborations have further enriched his work and contributed to advancements in their shared field.

Conclusion

Joaquim Gomes Da Costa Eulalio De Souza is a distinguished inventor whose work in neural network processing is shaping the future of technology. His innovative patent demonstrates his commitment to enhancing computational efficiency, making significant strides in the field.