Hierarchical User Behavior Recognition Using Sensor Data Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sensor-based user behavior recognition technologies lack accuracy in measuring movement and are unable to precisely recognize daily life activities such as eating, going to the toilet, or washing the face, due to limitations in data processing and analysis.
Innovation Solution
An apparatus and method that extracts unit data from sensor data, converts it into feature information using algorithms like Bayesian or SVM, and recognizes representative user behaviors by classifying unit behaviors into sequences or selecting significant behaviors based on predetermined criteria, utilizing a combination of sensors like accelerometers, GPS, and biosensors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If simple movement measurement is used for fitness tracking, then device complexity is reduced, but measurement precision deteriorates
Solution Approach 1:
The patent segments the behavior recognition process into multiple hierarchical levels: unit behavior recognition (basic movements), composite behavior recognition (combinations of unit behaviors), and representative behavior recognition (high-level activities). This segmentation allows the system to achieve high measurement precision through progressive analysis while keeping individual processing stages relatively simple and manageable.
Solution Approach 2:
The patent introduces temporal dimension by analyzing behavior sequences over time. Instead of treating each sensor data point independently, the system analyzes sequences of unit behaviors to identify composite behaviors and representative behaviors. This dimensional approach enables accurate recognition of complex daily activities without requiring overly complex single-step analysis.
2Measurement precision
If comprehensive sensor data analysis is performed to recognize daily life activities, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent divides the complex behavior recognition task into distinct hierarchical stages: extracting unit behaviors from sensor data, combining unit behaviors into composite behaviors, and identifying representative behaviors from composite behaviors. Each stage processes relatively simple information, avoiding the need for a single complex analysis system while achieving high overall measurement precision for daily life activities.
3Measurement precision
If hierarchical behavior recognition is implemented, then measurement precision is improved, but computational time increases
Solution Approach 1:
The patent performs preliminary processing by extracting unit behaviors from sensor data first, which are then reused in subsequent stages for composite and representative behavior recognition. This preliminary extraction avoids redundant processing in later stages, reducing overall computational time while maintaining the measurement precision benefits of hierarchical analysis.
Solution Approach 2:
The system focuses computational resources on extracting and analyzing the most significant behaviors at each hierarchical level. Rather than processing all possible behavior combinations equally, the patent identifies and processes only the relevant unit behaviors needed for recognizing representative behaviors, reducing unnecessary computational overhead while maintaining high measurement precision.
Data Source
AI summary
An apparatus for recognizing a representative user behavior includes a unit-data extracting unit configured to extract at least one unit data from sensor data, a feature-information extracting unit configured to extract feature information from each of the at least one unit data, a unit-behavior recognizing unit configured to recognize a respective unit behavior for each of the at least one unit data based on the feature information, and a representative-behavior recognizing unit configured to recognize at least one representative behavior based on the respective unit behavior recognized for each of the at least one unit data.