Classification Model Parameter Optimization via Simulated Signals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing methods for determining optimal acquisition parameters for classification models, such as sampling frequency and amount of data, are time-consuming and costly, involving trial and error with multiple acquisitions, and do not guarantee optimal performance.
Innovation Solution
A method that involves obtaining initial time-series signals with initial acquisition parameters, creating simulated signals with varied parameters, assessing test classification models, and selecting optimal acquisition parameters based on performance analysis to create a final classification model efficiently.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If trial and error acquisitions are performed at different sampling frequencies and amounts of data to find optimal classification model parameters, then the classification model performance can be optimized, but the process becomes time-consuming and costly
Solution Approach 1:
The patent applies preliminary action by using simulated time-series signals to pre-assess the performance of test classification models with different acquisition parameters before actual acquisitions are performed. This allows the optimal sampling frequency and data amount to be determined in advance, avoiding time-consuming trial and error acquisitions.
Solution Approach 2:
The patent creates simulated time-series signals that copy the characteristics of real signals but can be generated rapidly without physical acquisition. These simulated signals are used to evaluate multiple classification model configurations, replacing the need for multiple actual signal acquisitions and significantly reducing the time and cost required for parameter optimization.
2Reliability
If multiple acquisitions are performed with different acquisition parameters to test classification model performance, then optimal parameters can be determined, but the cost increases
Solution Approach 1:
The patent uses simulated time-series signals as copies of real signals to evaluate classification models. These simulated signals can be generated computationally without the energy cost of actual sensor acquisitions, processing, and storage, thereby determining optimal acquisition parameters at a fraction of the cost while maintaining the ability to assess model performance accurately.
Solution Approach 2:
By performing preliminary assessments using simulated signals and test classification models, the patent identifies the optimal acquisition parameters before committing resources to actual acquisitions. This preliminary action prevents wasteful spending on multiple costly acquisition campaigns with suboptimal parameters.
3Measurement precision
If larger amounts of data are acquired at higher sampling frequencies to improve classification model accuracy, then the model performance increases, but the energy consumption and memory occupation increase
Solution Approach 1:
The patent uses simulated time-series signals to preliminarily determine the optimal sampling frequency and data amount required for achieving satisfactory classification accuracy. This allows the system to identify the minimum necessary acquisition parameters in advance, avoiding the energy waste associated with acquiring excessive data at unnecessarily high sampling rates.
Data Source
AI summary
A method for creating a classification model includes: obtaining at least one group of initial time-series signals associated with at least one initial acquisition parameter, creating at least one group of simulated time-series signals from the at least one group of initial time-series signals, creating various test classification models, from groups of initial or simulated time-series signals, assessing the performances of each test classification model, obtaining at least one group of final time-series signals associated with at least one final acquisition parameter, and creating the classification model.