Sensory Characterization Modeling With Context-Aware Intensity Ranking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for characterizing sensory characteristics, such as flavor, are subjective and inaccurate due to variability among human panelists, inter-sample fatigue, and complex component interactions, requiring extensive human data for each compound-concentration permutation.
Innovation Solution
A method that ranks perceived sensory intensity instead of using conventional scales, incorporating sample context and predicting olfactory and gustation compositions to enhance accuracy, using machine learning models to predict sensory characterizations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional sensory characterization methods using human panelists and subjective scales are used, then sensory data can be collected, but measurement precision and reliability deteriorate due to panelist variability and inter-sample fatigue
Solution Approach 1:
The patent replaces the mechanical system of human sensory evaluation with an in silico computational model. The trained machine learning model predicts sensory characterizations by processing chemical composition data, eliminating human panelists entirely. This substitution resolves the reliability issues caused by panelist variability and fatigue while maintaining measurement precision through consistent algorithmic processing.
Solution Approach 2:
The patent creates a virtual copy of the sensory evaluation process through a trained computational model. Instead of repeatedly using human panelists, the system trains a model on initial sensory data and then uses this digital twin to predict sensory characterizations for new samples. This copying approach preserves the sensory evaluation capability while eliminating the inconsistencies inherent in human assessment.
2Loss of information
If extensive human panelist data is collected for each compound-concentration permutation, then comprehensive sensory coverage is achieved, but productivity and time consumption worsen
Solution Approach 1:
The patent performs preliminary action by training the computational model on a comprehensive dataset of compound-concentration permutations beforehand. Once trained, the model can rapidly predict sensory characterizations for new samples without requiring additional human panelist evaluations. This preliminary training phase captures the necessary sensory information, enabling fast subsequent predictions and resolving the productivity bottleneck.
Solution Approach 2:
The system enables self-service by allowing the computational model to generate sensory characterizations autonomously without requiring human panelist intervention for each new sample. The model serves itself by processing chemical composition data and outputting predicted sensory profiles, dramatically increasing productivity while maintaining data completeness through the model's comprehensive training.
3Measurement precision
If complex component interactions are fully characterized through human paneling, then accurate sensory profiles are obtained, but device complexity and data processing requirements increase
Solution Approach 1:
The patent applies parameter changes by transforming the complex sensory evaluation problem into a computational learning task. The model learns to map chemical composition parameters to sensory characterization parameters through training data. This parameter transformation approach maintains measurement precision for complex interactions while managing data processing complexity through efficient machine learning algorithms rather than exhaustive human evaluation protocols.
Data Source
AI summary
In variants, a method for sensory characterization can include: determining attributes for a sample, collecting sensory data for the sample, determining a sensory characterization model, training the sensory characterization model based on the attributes and the sensory data, determining attributes for a test sample, and predicting a sensory characterization for the test sample using the sensory characterization model.