Classifier Fusion Module for Multi-Level User State Discrimination
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current operator state classification systems can only distinguish between two states, such as high and low workload, which limits their precision and adaptability, as they fail to provide the necessary higher resolution state tracking and adaptation selection required for next-generation adaptive systems, aiming for classification performance above 90% to ensure user acceptance and trust.
Innovation Solution
A system and method that combines the outputs of multiple two-state classifiers using a classifier fusion module to classify user states into more than two levels, allowing for the distinction of at least three different classification states by integrating data from sensors such as EEG, ECG, and other physiological and contextual sensors, and applying algorithms like linear and nonlinear classifiers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple two-state classifiers are fused to increase state level discrimination, then measurement precision improves, but device complexity increases
Solution Approach 1:
The classification task is segmented into multiple independent two-state classifiers, each handling a specific dimension (e.g., workload, alertness). Each classifier processes sensor data separately and produces a binary output, which are then combined through fusion to achieve multi-level state discrimination without requiring a single complex classifier
Solution Approach 2:
Multiple two-state classifier outputs are merged through a fusion mechanism (e.g., logical operations, voting, or weighted combination) to produce a final multi-level classification. This combining approach leverages the simplicity of individual two-state classifiers while achieving the precision of multi-level classification
2Adaptability or versatility
If more than two classification states are implemented, then adaptability improves, but reliability decreases
Solution Approach 1:
The classification problem is divided into multiple binary classification tasks, each with high reliability (above 90% as stated in the patent). By segmenting the complex multi-level classification into simpler binary decisions, each segment maintains high reliability while collectively providing diverse adaptation options
Solution Approach 2:
The fusion module acts as an intermediary that combines reliable binary classifier outputs into a multi-level classification. This intermediary layer translates the high-reliability binary decisions into actionable multi-level states without directly compromising the reliability of the underlying classifiers
Data Source
AI summary
A system and method for providing more than two levels of classification distinction of a user state are provided. The first and second general states of a user are sensed. The first general state is classified as either a first state or a second state, and the second general state is classified as either a third state or a fourth state. The user state of the user is then classified as one of at least three different classification states.