Half a century of neural networks -- What's Next?

McClelland, J.

This talk will consider the past, present, and future of neural networks, beginning with the pioneering work of Rosenblatt in the 1950's. This history has been marked by controversy and by fits and starts in the development of neural network ideas. I will argue that real progress has occurred and that the core principles recognized by Rosenblatt and other pioneers lie at the heart of what has been achieved. I will also point to developments in neuroscience, methods of probabilistic computation, and computing technology that have contributed to this progress, and that make the period ahead seem full of promise.