Feature Quantity Selection for Stress Estimation Models

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current feature quantity selection methods for stress level estimation models do not consider the properties of various feature quantities, leading to suboptimal performance and increased costs due to the 'curse of dimensionality' issue.

Innovation Solution

An information processing apparatus and method that selects feature quantities based on utility evaluation results for each modality, generating a feature set and then verifying estimation accuracy to select a combination of feature quantities for machine learning, thereby improving the feature quantity selection process for stress level estimation models.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If many feature quantities are collected for stress estimation, then the comprehensiveness of the estimation model is improved, but the learning accuracy deteriorates due to the curse of dimensionality when data samples are limited

Engineering Contradiction:

Improvecomprehensiveness of estimation modelVSAvoidlearning accuracy

Core Design Contradiction:

Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The patent extracts and selects only the most relevant feature quantities from a large set of candidate features using utility evaluation and verification processes. This extraction principle resolves the contradiction by removing unnecessary features that contribute to the curse of dimensionality while retaining the essential features needed for comprehensive stress estimation.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent changes the parameter of feature quantity selection from a static comprehensive approach to a dynamic optimized approach. By introducing utility evaluation results and verification-based selection, the system adapts the feature set size and composition to match the available data samples, thereby resolving the contradiction between comprehensiveness and learning accuracy.

Inventive Principle:

Principle #35Parameter changes

2Ease of operation

If general feature quantity selection algorithms are used without considering modality properties, then the selection process is simplified, but the estimation accuracy deteriorates

Engineering Contradiction:

Improvesimplicity of selection processVSAvoidestimation accuracy

Core Design Contradiction:

Ease of operationVSMeasurement precision

Solution Approach 1:

The patent segments the feature quantity selection process into distinct stages: utility evaluation based on modality properties, generation of candidate feature sets, and verification-based selection. This segmentation allows the system to incorporate complex modality-specific considerations without overwhelming the overall process, resolving the contradiction between simplicity and accuracy.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary utility evaluation of feature quantities based on modality properties before the actual selection process. This preliminary action prepares and organizes the feature candidates in advance, making the subsequent selection process more efficient and accurate while maintaining operational simplicity.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS20240104430A1Information processing apparatus, feature quantity selection method, training data generation method, estimation model generation method, stress level estimation method, and storage medium

Publication Date: 2024.03.28 NEC CORP

AI summary

In order to improve a feature quantity selection method for machine learning of a stress level estimation model, an information processing apparatus (1) includes: a first selection section (11) that generates a feature set by selecting a feature quantity corresponding to each of a plurality of modalities from among a plurality of feature quantities; and a second selection section (12) that selects, based on a result of verifying estimation accuracy, a combination of feature quantities for use in the machine learning, the verification being carried out by applying combinations of feature quantities included in the feature set to the machine learning of the estimation model.