neural network hardware design

Adaptive Computation (ADC) is a spin-off company from Caltech developing transformative perception and cognition tools for intelligent machines. Its biomimetic design is based on the mammalian extended visual pathway (EViP), combining unsupervised learning for dynamic perception with self-evolving supervised learning. ADC is designing a hybrid high-speed correlator/neural network EViP chip and is developing a neural compiler and machine learning language in software, plus a compact low-power high-speed neural processor in hardware.

Read more

machine learning software developmentneural processor developmentsupervised learningUnsupervised Learning