Learning to see

Abstract

The mind cannot make sense of the visual world out of raw image data alone. In an approach to visual processing known as learning from examples, computational neural networks and physiological studies suggest how neurons and machines adapt to novel images on the basis of past experience.