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

Method and device for performing adversarial training by using Internet pictures based on active learning

An active learning, Internet technology, applied in the field of robust machine learning, can solve the problems of hindering the application of confrontation training and high time cost, and achieve the effect of shortening training time, efficient training, and reducing time cost

Pending Publication Date: 2020-09-29
TIANJIN UNIV
View PDF6 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, it has a serious flaw - the time cost is too high
If adversarial training is introduced into the AI ​​intelligent image review system, the high time cost is immeasurable in the face of a large amount of data, and it is also impractical to use all data for training, which will seriously hinder the application of adversarial training in real scenarios

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
  • Method and device for performing adversarial training by using Internet pictures based on active learning
  • Method and device for performing adversarial training by using Internet pictures based on active learning
  • Method and device for performing adversarial training by using Internet pictures based on active learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0035] A method for adversarial training using Internet images based on active learning, see figure 1 with figure 2 , the method includes:

[0036] 1. Use an active learning strategy to select some image samples for training, and use the trained model as a discriminator to judge and select the image samples to be selected;

[0037] Through the above processing, redundant data is avoided to generate adversarial samples, and unnecessary calculations are reduced. It can significantly improve the training speed when the robustness improvement effect is guaranteed to be high. Reduce the excessive dependence on the amount of data. When faced with massive image data, the model can learn independently and efficiently, which is more intelligent and greatly reduces the cost of manual labeling.

[0038] 2. Select image sample data directly from the original image data set, use the selected image sample data to generate adversarial samples, and then perform adversarial training;

[0...

Embodiment 2

[0053] Combine below Figure 1 to Figure 4 The scheme in Example 1 is further introduced, see the following description for details:

[0054] 1. Using massive picture data on the Internet as a data pool, according to the picture identification and review tasks of the AI ​​intelligent picture review system, multiple types of illegal pictures can be selected as the picture training set first, and the selected picture data set can be used as the initial training set X;

[0055] Wherein, the AI ​​intelligent picture review system is well known to those skilled in the art, and the embodiment of the present invention does not repeat it here.

[0056] 2. Construct the convolutional neural network model of the AI ​​intelligent picture review system, and use the above-mentioned selected initial training set X to train the neural network model to obtain the initial convolutional neural network model. The model structure is as follows image 3 shown;

[0057] The AI ​​intelligent pictu...

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 discloses a method and device for adversarial training by using Internet pictures based on active learning. The method comprises the steps of: selecting a part of picture samples for training through employing an active learning strategy, and employing a trained model as a discriminator to judge and select a to-be-selected picture sample; and selecting picture sample data from an original picture data set, generating adversarial samples by utilizing the selected picture sample data, and then performing adversarial training. The adversarial sample selection strategy problem is improved, the uncertainty of samples is measured based on the probability, representative samples are efficiently selected through the strategy, and then adversarial samples are generated for learning; and a classification maximum probability value is selected as an index for measuring the uncertainty of the samples. The device comprises a memory and a processor, and the processor implements the steps of the method when executing a program.

Description

technical field [0001] The present invention relates to the field of robust machine learning, in particular to a method and device for adversarial training based on active learning using Internet pictures, which selectively utilizes adversarial samples to implement adversarial training and improve the training rate. Background technique [0002] With the continuous improvement of hardware conditions, the processing of massive data has been realized, and deep learning has developed rapidly. The deep network model is inspired by biological neural networks, which can learn different data features from a large number of data samples, and then perform tasks such as classification and regression. It is widely used in computer vision and natural language processing. In areas such as image recognition, its performance is superior to traditional classifiers. However, the neural network has complex and unrobust characteristics, and it will show some unpredictable results when the da...

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
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/62G06N3/04
CPCG06N3/045G06F18/214
Inventor 蒋正晖陶文源姚雯夏宇峰
Owner TIANJIN UNIV
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