Andrea LiuUniversity of Pennsylvania
“AI without a Computer”
Artificial neural networks are everywhere now because they are so useful. They do everything from predicting the weather to writing essays and translating them into Urdu. Their abilities have exploded during the last dozen years because they have grown far bigger. Unfortunately, though, bigger neural networks require more energy to train and run. Our brains use only about a fifth of the calories we eat to learn and perform a far greater variety of tasks. How do the networks of neurons in our brains manage to do all this at low energy cost? The answer is that they doesn’t use a computer—they learn on their own. Neurons update their connections without knowing what all the other neurons are doing. We have developed an approach to learning that shares this key property. Our approach does not mimic the brain but exploits physics to learn and perform tasks for us. Using this approach, our collaborators have built physical systems that learn and perform AI tasks on their own. Our work establishes a new paradigm for scalable, energy-efficient learning.