We target the problem of closed-loop learning of control policies that map visual percepts to continuous actions. Our algorithm, called Reinforcement Learning of Joint Classes (RLJ...
— Controlled electrical stimulation of the retina can result in visual percepts in blind patients. In contrast to the over 100,000,000 photoreceptors in a healthy retina, even hu...
Wentai Liu, Wolfgang Fink, Mark A. Tarbell, Mohana...