Field Programmable Field Arrays distinguish themselves from other integrated circuits, by their reconfigurability. This, along with their parallelism opens possibilities for implementing artificial neural networks with a high efficiency regarding cost and energy use. Research conducted on the implementation of Neural Network Architectures like VGG16 and AlexaNet in FPGAs can result in systems that are faster and more power efficient than GPU implementations.
Информационные технологии и математическое моделирование (ИТММ-2021) : материалы XX Международной конференции имени А. Ф. Терпугова, 1−5 декабря 2021 г.. Томск, 2022. С. 310-313