Guidance-type policy search reinforcement learning algorithm

A technology of reinforcement learning and strategy search, applied in the field of machine learning, can solve the problems of undiscovered patents, literature reports, etc., and achieve the effect of solving the large demand for samples, accurate strategy search, and reducing the number of samples

Inactive Publication Date: 2016-09-21
TIANJIN UNIVERSITY OF SCIENCE AND TECHNOLOGY
View PDF0 Cites 20 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] Through searching, no patents and literature

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
  • Guidance-type policy search reinforcement learning algorithm
  • Guidance-type policy search reinforcement learning algorithm
  • Guidance-type policy search reinforcement learning algorithm

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0025] The present invention will be further described in detail below in conjunction with the accompanying drawings and through specific embodiments. The following embodiments are only descriptive, not restrictive, and cannot limit the protection scope of the present invention.

[0026] A guided strategy search reinforcement learning algorithm, first selects high-quality learning samples according to the definition of guided learning samples, then uses the selected samples to perform gradient estimation on the objective function constructed in the present invention, and updates parameters according to the policy update principle until convergence . Specific steps are as follows:

[0027] (1) Sample collection: Under the framework of the Markov decision process, the agent is in the current state s, chooses an action a according to the current policy function π(a|s, θ), then transfers to the state s′, and receives an immediate reward r(s, a, s'). The agent collects state, act...

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 relates to a guided strategy search reinforcement learning algorithm. Firstly, a guided learning sample is selected, and then the selected sample is used to perform gradient estimation on an objective function, and parameters are updated according to a policy update principle until convergence. The invention greatly reduces the problem of lowering algorithm stability performance and convergence rate due to the use of important sampling technology by reconstructing the objective function. The invention defines a guided high-quality learning sample for reinforcement learning, through the use of the guided learning sample, the policy search can be performed more accurately, thereby avoiding a local optimum in a bad situation.

Description

technical field [0001] The invention belongs to the field of machine learning, and mainly relates to reinforcement learning algorithms, in particular to a strategy search reinforcement learning algorithm for continuous state action space. Background technique [0002] Machine learning is one of the core research fields of artificial intelligence, and its research motivation is to make the computer system have human learning ability so as to realize artificial intelligence. Reinforcement learning, as an important learning method in the field of machine learning, has been widely used in games, robots, scheduling systems, intelligent dialogue systems, storage systems, intelligent power generation control, intelligent transportation systems, unmanned vehicles, and aerospace systems. . Reinforcement learning is a continuous decision-making process. It does not require prior knowledge. Instead, the agent obtains knowledge through continuous interaction with the environment, and a...

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): G06F17/15G06N5/02
CPCG06F17/15G06N5/022
Inventor 赵婷婷杨巨成赵希陈亚瑞房珊珊
Owner TIANJIN UNIVERSITY OF SCIENCE AND TECHNOLOGY
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap