Research On Sparse Prediction Coding Based On Jacobian Constraints

Sparse coding is widely used in the field of signal processing.In recent years,a large number of algorithms have been proposed based on different priors or constraints.However,these methods need to be implemented through the corresponding iterative process,until the algorithm converges,the corresponding sparse representation can be obtained,In this paper,based on the idea of approximating the fixed point mapping in sparse coding,it is proved theoretically that the mapping in sparse coding is differentiable and has Jacobian matrix,and the predictive sparse coding representation obtained by sparse coding mapping is sparse,Furthermore,a learning method of predictive coding based on Jacobian constraint is proposed,and the neural network structure of predictive sparse coding in this method is defined.Finally,the effectiveness of this method is verified by simulation experiments,and the performance of this method is better than other traditional sparse coding methods in classification tasks.