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

Unified interactive robot multi-task learning method based on attention mechanism

A multi-task learning and attention technology, applied in the field of artificial intelligence, can solve problems such as unfriendliness, high reasoning delay, negative impact of real-time interactive experience, etc., and achieve the effect of interactive intelligence

Pending Publication Date: 2022-05-13
山东新一代信息产业技术研究院有限公司
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Multiple model deployments will increase the memory overhead of the robot computing platform, and generate more calculations and higher inference latency, which is quite unfriendly to the robot computing platform with limited resources, and has a negative impact on the experience of real-time interaction

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
  • Unified interactive robot multi-task learning method based on attention mechanism

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0035] refer to figure 1 , a unified attention mechanism-based multi-task learning method for interactive robots, including the following steps:

[0036] S1. Use deep learning programming frameworks such as TensorFlow or PyTorch to build a model architecture. The basic operation of the attention mechanism is matrix multiplication. The specific implementation method is as follows:

[0037] Mark the feature map output by the feature extractor as F1 and convert it into a feature matrix in spatial order and record it as M1. The dimension of M1 is C rows and W*H columns, where C is the number of channels of F1, and W and H are F1 respectively. width and height;

[0038] In the same way, the feature maps output by the three local networks are recorded as F2, F3, and F4, respectively, and converted into three feature matrices, which are respectively recorded as M2, M3, and M4;

[0039] The number of channels of the feature map output by the feature extractor and the local network i...

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 provides a unified interactive robot multi-task learning method based on an attention mechanism. A model can recognize facial expressions, age distribution and gender information in an image. A unified interactive robot multi-task learning method based on an attention mechanism comprises the following steps: S1, carrying out multi-task learning by a unified deep convolutional neural network model based on the attention mechanism, and identifying facial expressions and evaluating age and gender at the same time; and S2, training and optimizing the model in the step S1: 1) setting a loss function for the model realized in the step 1), 2) inputting a data set and training the model, 3) stopping training when the model convergence or the accuracy reaches the application requirement, and S3, deploying the trained model in the step S2 to the robot.

Description

Technical field [0001] The invention relates to a unified multi-task learning method for interactive robots based on attention mechanism, and belongs to the technical field of artificial intelligence. Background technique [0002] With the application of deep learning technology in robot human-computer interaction, robots are becoming more and more intelligent and humane. In the process of emotional interaction between robots and people, the primary problem is that the robot first needs to identify the basic information of the current interacting person, such as the interacting person's emotional state, gender, age and other information, and then choose different interaction strategies to achieve a more intelligent and humane purpose, thereby enhancing the value of the robot. In order to achieve the purpose of sensing the emotions, gender, age and other information of the interacting person without preset information, it is often necessary to install a camera in the robot t...

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): G06V40/16G06V10/82G06N3/04G06N3/08G06N5/04
CPCG06N3/08G06N5/04G06N3/045
Inventor 高岩郝虹王雯哲王建华高明尹青山
Owner 山东新一代信息产业技术研究院有限公司
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