Biological Sensor Dependency Features for Time-Series Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing machine learning systems struggle to effectively process multivariate time series data from biological sensors due to the lack of efficient feature extraction methods that consider the statistical dependencies between sensors, leading to suboptimal prediction performance.
Innovation Solution
A method that calculates statistical dependencies between samples from multiple biological sensors, generates a relationship matrix, and transforms it into feature vectors to capture these dependencies, enhancing the training of machine learning models for improved prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If traditional feature extraction methods are used, then processing speed is improved, but prediction accuracy deteriorates because statistical dependencies between sensors are ignored
Solution Approach 1:
The patent performs preliminary calculation of statistical dependencies between sensors during the feature extraction phase, generating a relationship matrix before the main prediction task. This preliminary action captures inter-sensor correlations in advance, allowing the prediction model to utilize these pre-computed relationships without adding significant processing time during real-time operation, thus resolving the contradiction between processing speed and prediction accuracy.
2Measurement precision
If statistical dependencies between sensors are calculated and incorporated, then prediction accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent segments the computational task by dividing sensor data into multiple channels and calculating statistical dependencies separately for each channel pair. This segmentation allows parallel processing of dependency calculations and enables the use of efficient algorithms for each segment, reducing overall computational complexity while maintaining the ability to capture inter-sensor relationships for improved prediction accuracy.
Solution Approach 2:
The patent transforms the relationship matrix derived from statistical dependencies into a different dimensional representation that is more suitable for machine learning model input. This dimensionality transformation reduces the complexity of the data structure while preserving the essential statistical relationship information, allowing accurate predictions with lower computational burden.
3Loss of information
If multiple biological sensors are used to sense biological functions, then measurement comprehensiveness is improved, but data processing requirements increase
Solution Approach 1:
The patent merges the information from multiple biological sensors by calculating statistical dependencies between them and combining these relationships into a unified relationship matrix. This merging process integrates data from multiple sensors in a compact representation that captures their interrelationships, reducing the overall data processing requirements while maintaining comprehensive measurement information for accurate predictions.
Data Source
AI summary
A computer implemented method includes accessing a multivariate time series set of samples collected by multiple biological sensors sensing a first biological function over a first period of time, dividing the data set into windows, calculating statistical dependencies between the samples of the timeseries data collected by each sensor, generating a relationship matrix as a function of the statistical dependencies, and transforming the relationship matrix to generate a first feature vector for each window of time that captures the statistical dependencies amongst the sensors.