Binary neural network forward propagation framework suitable for mobile terminal

Abstract

The invention relates to a binary neural network forward propagation framework suitable for a mobile terminal. The binary neural network forward propagation framework comprises a model conversion module and a forward propagation module, wherein the forward propagation model comprises a data storage unit, an operation unit and an optimization unit; the data storage unit stores data in a data layoutmode of number, height, width and channel number and compresses the data; wherein the arithmetic unit is a GPU arithmetic unit and merges the operation layers of the binarized neural network; the optimization unit balances the thread read-write data amount and the calculated amount in the arithmetic unit. Compared with the traditional neural network framework, the scheme of forward propagation onthe mobile phone is provided, the storage occupation is reduced, the operation speed is improved, the energy consumption ratio of the binary neural network running on the mobile phone is improved byutilizing the GPU, and the binary neural network can be efficiently run on the mobile terminal.

Description

technical field [0001] The present invention relates to the field of neural network frameworks, and more particularly, relates to a binary neural network forward propagation framework suitable for mobile terminals. Background technique [0002] Artificial Neural Networks (ANNs for short), also referred to as Neural Networks (NNs) or Connection Model, is a kind of algorithmic mathematics that imitates the behavior characteristics of animal neural networks and performs distributed parallel information processing. Model. This kind of network depends on the complexity of the system, and achieves the purpose of processing information by adjusting the interconnection relationship between a large number of internal nodes. [0003] When applying an artificial neural network, it is generally divided into four steps: 1. Train the corresponding neural network according to the requirements; 2. Adjust and optimize the trained neural network to obtain a suitable model for deployment; 3. ...

AI Technical Summary

Problems solved by technology

[0005] The present invention provides a binary neural network forward propagation framework suitable for mobile terminals in order to overcome the problems of low computational efficiency, high energy consumption and slow speed caused by the mobile terminal only using the CPU to perform artificial neural network forward reasoning in the prior art. The framework runs in the GPU of the mobile terminal and is optimized through computing to improve computing efficiency, reduce energy consumption and increase speed

Images