Big data processing method based on deep learning model satisfying K order sparsity constraint

摘要

The present invention proposes a big data processing method based on a deep learning model satisfying the K order sparsity constraint. This method constructs a deep learning model satisfying the K order sparsity constraint using an unlabeled training sample by a gradation pruning method, wherein the K order sparsity constraint is a node K order sparsity constraint and a hierarchy K Wherein said training sample after update is input to a depth learning model that satisfies said K order sparsity constraint, optimizing a weight parameter of each layer of the model to satisfy a K order sparsity constraint, Step 2) of acquiring a deep learning model that has been subjected to the K order sparsity constraint, and a step 3) of inputting the big data to be processed to the optimized deep learning model satisfying the K order sparsity constraint and processing, finally outputting the processing result ,including. According to the method of the present invention, it is possible to reduce the difficulty of processing big data and to improve the processing speed of big data.(FIG.