Temporal Data Augmentation Using Wavelet Heat Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing predictive data analysis processes are unable to accurately process and interpret primary time-based physiological signals, leading to inaccurate and unreliable inferences based on secondary factors that may be weakly correlated with predictive insights.
Innovation Solution
A machine-learning based temporal classification model is introduced, which performs a multi-stage predictive classification process by generating a multi-channel data structure from time-based data, emphasizing temporal features, and using wavelet transforms to create heat maps, allowing for improved prediction accuracy through multiple, potentially differently weighted channels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing predictive data analysis processes use secondary factors for predictions, then the processes can operate without directly processing primary time-based signals, but the prediction accuracy deteriorates due to weak correlation with predictive insights
Solution Approach 1:
The patent segments the prediction process into multiple stages: raw signal acquisition, feature extraction, feature selection, and prediction. This segmentation allows direct processing of primary time-based signals while managing complexity through structured workflow decomposition, resolving the contradiction between using primary signals for accuracy and maintaining process simplicity.
Solution Approach 2:
The patent transforms one-dimensional time-based raw signals into multi-dimensional feature representations through feature extraction. This dimensional transformation enables the system to capture complex temporal patterns and relationships that improve prediction accuracy while providing a structured framework for processing that manages computational complexity.
2Measurement precision
If primary time-based physiological signals are directly processed, then prediction accuracy improves, but the processing complexity and computational requirements increase
Solution Approach 1:
The patent applies preliminary feature extraction and feature selection operations on raw time-based signals before the actual prediction process. This preliminary action transforms complex raw signals into simplified, informative features that retain essential temporal patterns, thereby improving signal interpretation accuracy while reducing the computational burden on subsequent prediction models.
Solution Approach 2:
The patent introduces feature extraction and feature selection as intermediary processing steps between raw signal acquisition and prediction. These intermediaries transform complex primary signals into simplified representations that maintain interpretability and predictive power, resolving the contradiction between direct signal processing for accuracy and complexity management.
3Reliability
If multi-stage processing with feature extraction and selection is implemented, then prediction reliability improves, but the processing time and computational resources increase
Solution Approach 1:
The patent performs feature extraction and feature selection as preliminary actions before the main prediction task. By preparing optimized feature representations in advance, the system improves predictive classification reliability while enabling faster real-time inference, as the computationally intensive feature engineering is done once rather than repeatedly during prediction.
Solution Approach 2:
The patent transforms raw signals into features with optimized parameters that enhance predictive power while reducing dimensionality. This parameter transformation concentrates information into fewer, more meaningful features, improving reliability while reducing the computational time required for processing compared to using all raw signal data.
Data Source
AI summary
Various embodiments of the present disclosure disclose machine-learning based data augmentation and prediction techniques for generating predictive classifications based on temporal data. A machine-learning based model is provided that can receive an input data object associated with a plurality of predictive temporal parameters; determine augmented temporal data objects based on the predictive temporal parameters; generate predictive data representations for the input data object based on the predictive temporal parameters and the augmented temporal data objects; generate a multi-channel predictive data representation based on the predictive data representations for the input data object; and generate a predictive classification for the input data object based on the multi-channel predictive data representation.