Coding distributed computing method and device, computer equipment and storage medium

Abstract

The invention relates to a coding distributed computing method and device, computer equipment and a storage medium, and the method comprises the steps: obtaining a matrix and a vector which need to becoded when a code obtains a combination characteristic; k equal division is carried out on a matrix needing to be coded in the vertical direction to obtain N data packets, and N is larger than or equal to K; respectively transposing the N data packets to obtain N intermediate data packets; shifting the N intermediate data packets according to a preset shift matrix and adding the N intermediate data packets to obtain N composite matrixes; carrying out transposition operation on the vector and then respectively calculating the vector and the N synthesis matrixes to obtain combined characteristics; and outputting the combined characteristics to a terminal for displaying at the terminal. According to the invention, the additional calculation burden introduced during encoding and decoding of the existing CDC framework is reduced, and the encoding and decoding processes are accelerated, so that the whole training process is accelerated.

Description

Technical field [0001] The present invention relates to computers, and more specifically to coding distributed computing methods, devices, computer equipment and storage media. Background technique [0002] The terminals of the Internet of Things often need to do large-scale computing tasks such as large-scale matrix-vector multiplication. In recent years, edge devices have increased substantially, and all edge devices have certain computing capabilities and can be regarded as working nodes of edge computing. Edge computing can calculate and process data at or near the source of data generation. Because edge devices often cannot bear heavy computing burdens, such as weak computing power and fast battery energy consumption, and data analysis and data processing are impaired by artificial intelligence. For many machine learning algorithms, such as gradient descent and logistic regression, the entire training process involves large-scale matrix-vector multiplication. In addition, a...

AI Technical Summary

Problems solved by technology

N working nodes respectively calculate C 1 x,C 2 x,...,C N x, when the calculation is completed, the calculation result will be returned to the master node. The master node can recover Ax after receiving the calculation result of any K of the N nodes. The existing CDC based on linear combination introduces a large number of multiplication and division methods in the calculation process , the LC-based CDC encoding process and decoding process are more computationally burdensome, increasing the overall training time

Images