We discuss Neuromorphic Computing, a term originated by Carver Mead at Caltech who pioneered work on a biologically-inspired VLSI Retina. The increased understanding of neuro-biological neurons and neural pathways along with theoretical advances has led to a variety of analog and digital organizations, with distinct performance, cost and energy characteristics. The widespread success of Deep Neural Network applications in myriad applications ranging from image processing, language transcription, autonomous vehicles and such has spurred activity in high performance computing architectures hardware to support their computational and energy requirements. However, there is still an enormous gap between such architectures and the accuracy and efficiency of neuronal and cortical systems in nature. This creates an opportunity to explore novel, biology-inspired, hardware-software architectures that use novel materials, more biologically-accurate neuron models, and analog computation on sparse signals. https://lmcad.ic.unicamp.br/news.html

--

neuromorphic computing