Study of the Grid Resource Neural Network Reservation Technology Based on SLA

Abstract

Reasonable resource reservation is a base for providing a good service in the Grid computation. The technology using neural network resource reservation based on service level agreement is proposed in this paper. A framework for the resource reservation using neural network decision system is built and a resource reservation decision control strategy based on neural network is researched. According to the state of grid services, with resource reservation neural network prediction categories for decision-making basis, the algorithm is described and analyzed by the examples, and is verified by the experiments. The simulation results show that the proposed grid resources reservation decision technology is able to predict the reasonable grid resource reservation category value. Combining with the current state of the grid service decision-making out grid can actually offer service level, and the service request has the high response efficiency and the components of the gird service request and response is of certain practical reference value.

Human Motion Target Recognition Using Convolutional Neural Network and Global Constraint Block Matching

“…CNN is a typical deep learning neural network [38], [39]. It consists of the convolution layer, pooling layer, normalized layer and Concat layer.…”