MIT Researchers, Working On Analog Deep Learning, Introduce A New Hardware Powered By Ultra-Fast Protonics And With Much Less Energy

The amount of time, effort, and resources needed to train increasingly complicated neural network models is soaring as more machine learning experiments are being done. In order to combat this, a brand-new branch of artificial intelligence called "analog deep learning" is on the rise. It promises faster processing with far less energy consumption. Like transistors are the essential components of digital computers, programmable resistors are the fundamental building blocks of analog deep learning. Researchers have developed a network of analog artificial "neurons" and "synapses" that can do calculations similarly to a digital neural network by repeatedly repeating arrays of programmable resistors in intricate layers.