Title: Predicting input spikes (and learning from the mistakes)

Bill Softky

DeepQ Technologies

Abstract:

When a sensory system constructs a model of the environment from its input, it might need to verify the model’s accuracy. One method of verification is multivariate time-series prediction: a good model could predict the near-future activity of its inputs, much as a good scientific theory predicts future data. Such a predicting model would require copious top- down connections to compare the predictions with the actual input. That feedback could further improve the models’ performance in two ways: by biasing internal activity toward expected patterns, and by generating specific error signals if predictions fail. A layered proof-of-concept network, with cortical features like spiking and local inhibition, was constructed to learn near-future predictions of a simple, moving “visual” stimulus. After unsupervised learning, the network contained units not only tuned to the obvious features of the stimulus like contour orientation and motion, but also to contour discontinuity (“end-stopping”) and illusory contours.