Wireless heterogeneous control computing system based on neural network

Abstract

The invention provides a wireless heterogeneous control computing system based on a neural network. The system includes a mobile intelligent terminal which is used for independently finishing requiredreal-time acquisition and communication, realizing complete neural network calculation and an edge calculation terminal with multi-system wireless communication, and providing neural network calculation guarantee without calculation burden and storage, synchronization and update of various data sources without storage burden for a host control terminal; the host control terminal which is used fordistinguishing, processing and bi-directionally executing the control instruction, obtaining data sent by the intelligent terminal through a network and participating in tail end neural network co-calculation to achieve indispensable dual-calculation guarantee. An asymmetric operator can be started at the host control terminal to ensure that the mobile intelligent terminal cannot bypass or copy peripheral neural network calculation dominated by a host controller and drive control equipment.

Description

technical field [0001] The present invention relates to the technical field of heterogeneous network computing, in particular to a neural network-based wireless heterogeneous control computing system. Background technique [0002] Human-machine remote control is one of the ways to facilitate operation and control. It plays an important role in the provision of various offline face-to-face exclusive services such as directional self-service, robots, identification access control, non-contact electronic locks, non-inductive recognition preference recommendations, etc. role. However, if these devices must have the computing tasks of a complete artificial intelligence neural network, these devices will inevitably need to collect some original customer data, which will bring certain requirements and constraints on user privacy, computing power consumption, and physical size requirements. This may limit the widespread application of artificial intelligence. [0003] For example,...

AI Technical Summary

Problems solved by technology

[0021] Patent document CN201711319211.1 discloses a full-scale distributed whole-brain simulation system based on brain-inspired spiking neural network, aiming to solve the lack of multiple-scale and multiple-scale modeling methods in brain-inspired simulation, as well as modeling and simulation coupling problem

Images