With the advent of big data,the cost of index recommendation(IR)increases exponentially,and the portability of IR model becomes an urgent problem to be solved.In this paper,a fine-grained classification model based on...With the advent of big data,the cost of index recommendation(IR)increases exponentially,and the portability of IR model becomes an urgent problem to be solved.In this paper,a fine-grained classification model based on multi-core convolution neural network(CNNIR)is proposed to implement the transferable IR model.Using the strong knowledge representation ability of convolution network,CNNIR achieves the effective knowledge representation from data and workload,which greatly improves the classification accuracy.In the test set,the accuracy of model classification reaches over 95%.CNNIR has good robustness which can perform well under a series of different learning rate settings.Through experiments on MongoDB,the indexes recommended by CNNIR is effective and transferable.展开更多
文摘With the advent of big data,the cost of index recommendation(IR)increases exponentially,and the portability of IR model becomes an urgent problem to be solved.In this paper,a fine-grained classification model based on multi-core convolution neural network(CNNIR)is proposed to implement the transferable IR model.Using the strong knowledge representation ability of convolution network,CNNIR achieves the effective knowledge representation from data and workload,which greatly improves the classification accuracy.In the test set,the accuracy of model classification reaches over 95%.CNNIR has good robustness which can perform well under a series of different learning rate settings.Through experiments on MongoDB,the indexes recommended by CNNIR is effective and transferable.
