Emotion Recognition Using SVM Hyperplane Weighting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current emotion recognition systems for robots are limited as they typically focus on either speech or facial expressions, failing to effectively utilize multiple input types and lacking the ability to accurately identify emotions with high reliability and speed.
Innovation Solution
An emotion recognition method that uses Support Vector Machine (SVM) to establish hyperplanes for emotion categorization, assigns weights to feature sets based on recognition reliability, and employs a Gaussian kernel function for space transformation to enhance accuracy and learning speed, allowing for the simultaneous use of facial and vocal data for improved emotion recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a robot uses single-mode emotion recognition (either speech or facial expression), then the system complexity is reduced, but the recognition accuracy and reliability deteriorate
Solution Approach 1:
The patent merges speech recognition and facial expression recognition into a unified emotion recognition system. Multiple feature extractors process different input modalities (speech features and facial features) separately, then their results are integrated through a common SVM classifier to produce final emotion recognition results, thereby improving reliability without excessive complexity increase
Solution Approach 2:
The system employs a universal SVM-based emotion recognition framework that can handle multiple input types (speech and facial expressions) through a single classification mechanism. The same SVM model structure processes different feature sets, providing multi-functional capability across different emotion recognition tasks
2Reliability
If multiple feature sets are used for emotion recognition, then the recognition accuracy is improved, but the calculation time and processing speed deteriorate
Solution Approach 1:
The system performs preliminary feature extraction and hyperplane establishment during an offline training phase using labeled data. Once the SVM model and hyperplanes are pre-computed, the online recognition phase only requires calculating distances to pre-established hyperplanes, significantly reducing real-time processing requirements while maintaining high accuracy
Solution Approach 2:
The patent transforms the recognition problem from a complex multi-feature classification task into a simpler geometric distance calculation problem by changing the parameter space representation. By establishing hyperplanes in feature space during training, the system converts multiple feature integration into a single distance metric computation during testing
3Reliability
If the system processes multiple input types (speech and facial data), then the recognition reliability is improved, but the device complexity and computational burden increase
Solution Approach 1:
The patent segments the emotion recognition system into independent modules: speech feature extractor, facial feature extractor, and SVM classifier. Each module processes its specific input type separately and produces feature representations that are then combined, allowing for modular complexity management and independent optimization of each processing stage
Data Source
AI summary
A method is disclosed in the present disclosure for recognizing emotion by setting different weights to at least of two kinds of unknown information, such as image and audio information, based on their recognition reliability respectively. The weights are determined by the distance between test data and hyperplane and the standard deviation of training data and normalized by the mean distance between training data and hyperplane, representing the classification reliability of different information. The method recognizes the emotion according to the unidentified information having higher weights while the at least two kinds of unidentified information have different result classified by the hyperplane and correcting wrong classification result of the other unidentified information so as to raise the accuracy while emotion recognition. Meanwhile, the present disclosure also provides a learning step with a characteristic of higher learning speed through an algorithm of iteration.