Decision Tree Filter Selection for Inertial Activity Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Establishing filters for decision trees to accurately distinguish between different movement activities using inertial sensors is challenging due to variations in human gait and the overlap of frequency peaks, making manual filter selection difficult and inaccurate.
Innovation Solution
A method involving peak-based and entropy-based frequency range selection to determine overlapping thresholds and select optimal filters for each class, followed by training a decision tree to enhance accuracy in identifying movement activities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual filter selection is used for decision trees, then the process is simple to implement, but the accuracy in detecting movement activities decreases due to variations in human gait and overlapping frequency peaks
Solution Approach 1:
The system performs self-service by automatically selecting optimal filters for decision trees without requiring manual intervention. The filter selection process autonomously analyzes sensor data characteristics and determines the best filters based on frequency peaks and entropy calculations, eliminating the need for expert manual configuration while improving detection accuracy across different human gaits.
Solution Approach 2:
The system dynamically changes filter parameters based on the characteristics of the sensor data being analyzed. By calculating frequency peaks and entropy values from actual movement data, the system adapts filter settings to match specific movement patterns, thereby improving accuracy for different human gaits without requiring manual parameter tuning.
2Measurement precision
If different filters are used for different people to improve detection accuracy, then the measurement precision improves, but the ease of operation decreases due to the difficulty of manually configuring filters for each individual
Solution Approach 1:
The system automatically adapts to each user's movement patterns by performing self-service filter selection. When a user performs a movement activity, the system analyzes the sensor data, identifies frequency peaks specific to that user's gait, and automatically configures appropriate filters without requiring manual input or configuration by the user.
Solution Approach 2:
The system uses feedback from sensor data to continuously optimize filter selection. By monitoring the characteristics of movement data and calculating entropy values, the system receives feedback about detection accuracy and automatically adjusts filter parameters to improve performance for each individual user over time.
3Productivity
If automated filter selection methods are implemented, then the productivity of developing decision trees improves, but the device complexity increases due to the need for frequency analysis and entropy calculation algorithms
Solution Approach 1:
The system performs preliminary actions by pre-calculating frequency peaks and entropy values from sensor data before the actual decision tree development process. This preliminary analysis prepares the optimal filter settings in advance, so when decision trees need to be developed or updated, the system can quickly apply pre-determined filters without performing complex calculations during the main development process.
Data Source
AI summary
Technological advancements are disclosed that utilize inertial sensor data for multiple classes to select a combination of filters to extract information though features to train a machine learning core decision tree. A determination is made whether the data for a class includes a frequency peak or dominating frequency that contains significant information about the class. In response to the data for the class including a frequency peak, a peak-based frequency range is determined. An entropy value is calculated for multiple frequency ranges in the data for the class. An entropy-based frequency range is selected from the multiple frequency ranges having a minimum entropy value. A frequency of interest is selected from the peak-based frequency range and the entropy-based frequency range for the class. A combination of filters is selected for each frequency of interest for each class and a decision tree is trained based on selected filter combination.