Unlock instant, AI-driven research and patent intelligence for your innovation.

An incremental cognitive development system and method integrating interactive reinforcement learning

A reinforcement learning and interactive technology, applied in the field of machine learning, can solve problems such as the implementation of open learning methods, and achieve the effect of improving the recognition accuracy.

Active Publication Date: 2021-03-30
SHANDONG UNIV
View PDF8 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

A limitation of this work is that it cannot be performed in an open-ended learning manner

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
  • An incremental cognitive development system and method integrating interactive reinforcement learning
  • An incremental cognitive development system and method integrating interactive reinforcement learning
  • An incremental cognitive development system and method integrating interactive reinforcement learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0069] In one or more embodiments, an incremental cognitive development system that integrates interactive reinforcement learning is disclosed, such as figure 1 As shown, it consists of a series of hierarchical self-organizing incremental neural networks. The learning process involves both bottom-up learning and top-down response. From a computational point of view, bidirectional structures have the ability to guide clustering and resolve conflicts autonomously, improving the accuracy of recognition. However, sometimes unavoidable misrepresentations can affect these strategies. To address this, we extend this cognitive architecture by fusing IRL and equipping some neurons with memory models. Objects and trainers are viewed as environments that provide perception and advice. Recognizing shapes, colors and names represent interactive actions. Thus, cognitive architectures are mainly involved in two processes: learning and practice, and these two processes can be performed in...

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 an incremental cognitive development system and method integrating interactive reinforcement learning, which introduces interactive reinforcement learning into a hierarchical self-organizing incremental neural network and can simultaneously learn object concepts and adjust learned knowledge through interaction with human beings. In order to realize the combination of the two algorithms, a memory model is configured for an individual neural network, and the model is designed to be an exponential function controlled by two forgetting factors to simulate the consolidationand forgetting process of human memory; an interactive reinforcement strategy is proposed for providing rewards or penalties and error correction is performed; these feedback acts on forgetting factors to strengthen or weaken the memory of the neurons, thereby preserving the correct representation while forgetting the erroneous representation. Experimental results show that human feedback can be effectively utilized, the learning effect is remarkably improved, and model redundancy is reduced.

Description

technical field [0001] The invention belongs to the technical field of machine learning, and in particular relates to an incremental cognitive development system and method that integrates interactive reinforcement learning. Background technique [0002] Recently, cognitive robots have received increasing attention in artificial intelligence. They can autonomously develop knowledge and skills to cope with various tasks and dynamic environments in human's daily life. The main challenge for current cognitive robots is how to quickly learn new objects they encounter and communicate properly with humans during human-robot interaction. Therefore, cognitive systems must have the ability to acquire knowledge online and correct errors in a timely manner based on human feedback. [0003] Typically, research on cognitive development in robots has focused on passive perception, such as seeing and listening. This approach, known as individualistic learning, can be used to learn conce...

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 Patents(China)
IPC IPC(8): G06N20/00G06F3/01G06N3/04
CPCG06F3/011G06N3/04G06N20/00
Inventor 马昕黄珂宋锐荣学文田新诚李贻斌
Owner SHANDONG UNIV