TRANSFORM FOLDING NEURONAL NETWORKS TO LEARN VISUAL SEQUENCES

摘要

A method, a computer readable medium, and a system for learning visual sequences using neural networks are disclosed. The method includes the steps of replacing a non-recurrent layer within a model of a trained convolutional neural network with a recurrent layer to generate a model of a visual sequences learning neural network, and feedforward weights for the non-recurring layer in input-to Transform borrowing weights of the recurrent layer to produce a transformed recurrent layer. The method also includes the steps of setting concurrent-to-hidden weights of the recurrent layer to initial values and processing video image data through the model of a visual sequences learning neural network to generate classification or regression output data.