An autonomous cognitive development system and method based on an incremental associated neural network and dynamic audio-visual fusion

Abstract

The invention discloses an autonomous cognitive development system and method based on an incremental association neural network and dynamic audio-visual fusion. The system comprises a sample layer, asymbol layer and an association layer three-layer network structure, Wherein the three-layer network structure comprises a visual path and an auditory path; In the visual path, a sample layer learnsthe original shape and color characteristics of an object respectively and performs autonomous clustering; The symbol layer is used for receiving the autonomous clustering result of the shape and color sample layer and abstracting the autonomous clustering result into a corresponding symbol; In the auditory path, a sample layer learns word vectors of names; The symbol layer is used for receiving the word vector category of the name and simplifying the word vector category into symbols; And the association layer establishes an association relationship between symbols in the visual path and theauditory path, and feeds back a response signal to the low-layer network according to the known association relationship. Based on the self-organizing neural network, the concepts of objects can be autonomously developed, and audio-visual fusion can be realized.

Description

technical field [0001] The invention belongs to the technical field of machine learning, and in particular relates to an autonomous cognitive development system and method based on incremental correlation neural network and dynamic audio-visual fusion. Background technique [0002] With more and more robots participating in human's daily life, cognitive development has become a hot spot in the field of intelligent robotics. Robots need to have the ability to recognize and understand when communicating with humans, so it is necessary to establish a common knowledge base with humans, such as the concept of objects. Robots are usually represented by human pre-designed internal knowledge, unable to adapt to unknown and dynamic environments. To solve this problem, robots need to have the ability to develop cognition autonomously like human babies. [0003] Using their own cognitive principles and parental guidance, human infants can rapidly develop representations of the world ...

AI Technical Summary

Problems solved by technology

[0007] Third, robots cannot develop cognition and establish a common knowledge base with humans through human-computer interaction

This method is computationally simple and consumes less memory, but is not suitable for restoring object representations

Images