Dynamic Object Recognition using Sparse Coded Three-way Conditional Restricted Boltzmann Machines

In this paper a novel framework capable of both accurate predictions and classifications of dynamic images is introduced. The proposed technique makes of use of a novel combination of sparse coding, a feature extraction algorithm, and three-way weight tensor conditional restricted Boltzmann machines, a form of deep learning. Experiments performed on both the prediction and classification of various images show the efficiency, accuracy, and effectiveness of the proposed technique.