Emotional Recognition Post-Processing via N-Dimensional Coordinates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Automated emotional recognition systems face inefficiencies due to the lack of uniformity in emotional representation diagrams, leading to significant alterations when transitioning between different diagrams, which hinders accurate emotional state classification and representation.
Innovation Solution
An automated emotional recognition system that includes an emotional state classifier and a post-processing function to combine emotional state indications from an input information stream, using a selected emotional representation diagram for optimized emotional state representation, allowing for context-specific and accurate emotional state recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If emotional state classifier is used to recognize emotions from input information stream, then automated emotional recognition can be achieved, but accuracy and reliability are reduced due to lack of uniformity in emotional representation diagrams
Solution Approach 1:
The patent transforms discrete emotional state indications into continuous n-dimensional coordinates within a unified emotional representation diagram. This parameter transformation allows for smooth transitions between emotional states and eliminates the discontinuity problem inherent in traditional classifiers that map to discrete categories. The n-dimensional space enables precise representation of emotional nuances while maintaining uniformity across different emotional states.
Solution Approach 2:
The patent introduces a post-processing function as an intermediary between the emotional state classifier and the final emotional recognition output. This intermediary component receives the classifier's discrete emotional state indications, transforms them into n-dimensional coordinates, and performs smoothing operations. This mediator resolves the conflict between the classifier's discrete output and the need for continuous, precise emotional representation.
2Device complexity
If emotional state classifier produces discrete emotional state indications, then classification is simplified, but smooth transitions and degrees of alikeness between emotions are lost
Solution Approach 1:
The patent elevates the emotional representation from discrete one-dimensional categories to continuous n-dimensional coordinates. This dimensional expansion allows emotional states to be represented as points in a continuous space, enabling smooth transitions through interpolation between adjacent emotional states. The n-dimensional space captures not only the type of emotion but also its intensity and nuanced variations.
Solution Approach 2:
The patent transforms the static, discrete emotional states into dynamic, continuous coordinates that can smoothly transition over time. The post-processing function applies smoothing operations that make the emotional trajectory continuous and differentiable, allowing for natural transitions between emotions rather than abrupt jumps between discrete categories.
3Adaptability or versatility
If different emotional representation diagrams are used, then various emotional perspectives can be captured, but significant alterations occur when transitioning between diagrams
Solution Approach 1:
The patent creates a universal n-dimensional emotional representation framework that can accommodate multiple emotional perspectives and diagram types. Instead of switching between different emotional representation diagrams, the system uses a single unified n-dimensional space that can represent various emotional models and perspectives consistently, eliminating the instability caused by transitions between different diagram systems.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An automated emotional recognition system (100) is provided. The emotional recognition system comprises an emotional state classifier (110) adapted to receive, during an operative phase, an input information stream (SD) with embedded information related to emotional states of a person, and to generate a succession of emotional state indications (SES) derived from said input information stream. The emotional recognition system further comprises a post-processing function (120), configured to receive at least two emotional state indications of said succession and, for each of said at least two emotional state indications, determine a corresponding emotional state representation in an emotional state representation system. The post-processing function is further configured to combine the emotional state representations of said at least two emotional state indications to obtain an output emotional state indication (OES).