Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Neural network architecture search method, neural network application method, equipment and storage medium

A technology of neural network and search method, which is applied in the fields of neural network application method, neural network architecture search, equipment and storage media, and can solve problems such as unsatisfactory performance, instability, and multiple resources

Inactive Publication Date: 2020-10-23
STATE GRID ZHEJIANG ELECTRIC POWER +1
View PDF5 Cites 33 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Although such a method can find a guarantee to find a better architecture, it needs to consume too many resources
Most search methods suffer from instability due to sensitivity of accuracy to random initialization and search architectures that sometimes yield unsatisfactory performance on other datasets or tasks

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Neural network architecture search method, neural network application method, equipment and storage medium
  • Neural network architecture search method, neural network application method, equipment and storage medium
  • Neural network architecture search method, neural network application method, equipment and storage medium

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0044] figure 1 A schematic flowchart of a neural network architecture search method in one embodiment of the present application is shown. As shown in the figure, the neural network architecture search method of this embodiment includes:

[0045] S11. Define a search space, the search space includes a preset number of nodes and various candidate operations between every two nodes, and use the search space as a network unit to be searched;

[0046] S12. Stack M network units to be searched in a predefined manner to obtain a neural network to be searched, wherein the initial value range of M is [4,6];

[0047] S13. Use the weight of each candidate operation in the neural network to be searched as a parameter to be optimized, use the search strategy of reinforcement learning and operation-level inactivation to search the network architecture of the neural network to be searched, and the reinforcement learning agent generates weight parameters, and Candidate operations with low...

Embodiment 3

[0083] image 3 It is a schematic flow chart of the neural network application method in another embodiment of the present application, such as image 3 As shown, the method mainly includes two parts: firstly, the target neural network is searched on the agent's dataset, and then the searched target neural network is assigned to image processing tasks, such as target detection and pedestrian re-identification. In the specific implementation of this embodiment, it can be subdivided into the following steps:

[0084] Step (1): Establish a search space, which is called a unit in this embodiment. Since the connection of the neural network conforms to the regulation of the directed acyclic graph, a directed acyclic graph composed of an ordered sequence of N nodes represents a unit in the network. Normally, a unit consists of 7 points, 2 input nodes, 4 intermediate nodes, and 1 output node. There are 8 candidate operations for each edge, and the corresponding weights are randomly ...

Embodiment 4

[0139] Embodiment four and embodiment five

[0140] In a third aspect, an embodiment of the present application provides a computer device, including a memory, a processor, and a computer program stored in the memory and operable on the processor. When the processor executes the computer program, the above-mentioned first aspect and its various The neural network-based image processing method described in the possible implementation manner.

[0141] In the fourth aspect, the embodiment of the present application provides a computer storage medium. The computer storage medium stores a computer program. When the computer program is executed by a processor, the neural network-based neural network described in the first aspect and its various possible implementations is realized. image processing method.

[0142] For the description of the third aspect, the fourth aspect and its various implementation modes in this application, you can refer to the detailed description in the fir...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention belongs to the technical field of deep learning, and particularly relates to a neural network architecture search method, a neural network application method, equipment and a storage medium. The method comprises the steps of searching a first neural network model through a reinforcement learning architecture search method, and performing network structure search on the first neural network model based on a preset jump connection number to obtain a second neural network model; and setting the structure of the second neural network model based on parameters of a pre-established evaluation network, and performing training on a test data set to obtain a trained neural network model. The neural network architecture in the method is higher in search speed, and the problem that network architecture search resources are consumed too much is effectively solved. The target image is processed through the neural network model obtained through training, various image processing taskscan be executed, and the recognition precision is higher during image processing.

Description

technical field [0001] The present application belongs to the technical field of deep learning, and specifically relates to a neural network architecture search method, a neural network application method, a device, and a storage medium. Background technique [0002] Deep learning has greatly promoted the research progress of computer vision. The core part of deep learning is the design and optimization of deep neural networks. Some popular models have been artificially designed to achieve the best artistic performance at that time. However, designing neural network architectures requires expertise and massive computing resources. This has changed with the advent of Neural Architecture Search (NAS), which aims to automatically discover powerful network structures and has achieved notable success in image recognition. In the early days of NAS, researchers focused on heuristic search methods, which sample architectures from a large search space and perform individual evaluat...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06N3/04G06N3/08
CPCG06N3/08G06N3/045
Inventor 张宏达郑斌王正国孙钢王舒颦林森陈思浩毛燕萍何韵蒋群张艺凡
Owner STATE GRID ZHEJIANG ELECTRIC POWER
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com