Sparse Coding Of Time-Varying Natural Images

We show how the principle of sparse coding may be applied to learn the forms of structure occurring in time-varying natural images. A sequence of images is described as a linear superposition of space-time functions , each of which is convolved with a time-varying coeecient signal. When a sparse, independent representation is sought over the coeecients, the basis functions that emerge are space-time inseparable functions that resemble the motion-selective receptive elds of cortical simple cells. Interestingly, the coeecients form a spike-like representation of moving images, and thus suggest an interpretation of spiking activity in the brain in terms of sparse coding in time.