Constrained Sample Selection for IoT Model Training
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
IoT devices face challenges in implementing machine learning due to limited memory and computation resources, exacerbated by machine learning algorithms that prioritize correctness over resource efficiency, leading to models that are not optimized for computation demand and memory footprint.
Innovation Solution
Selecting data samples for training and testing datasets that adequately represent the variability of the population, ensuring sufficient samples for accurate modeling while minimizing redundancy and hidden variables, to reduce model complexity and resource requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine learning algorithms optimize for correctness without regard for computation demand and memory footprint, then model accuracy is improved, but device resource consumption increases
Solution Approach 1:
The patent extracts and removes redundant and duplicate samples from the training dataset, keeping only the essential information needed for model training. This reduction in dataset size directly decreases computation demand while preserving model accuracy, as the remaining samples contain the critical patterns needed for correct classification.
Solution Approach 2:
The patent changes the parameter of dataset composition by selectively removing samples based on redundancy analysis. By transforming the dataset from a complete but redundant collection to a minimized essential set, the computation demand is reduced while maintaining the statistical properties necessary for accurate modeling.
2Measurement precision
If machine learning algorithms optimize for correctness without regard for computation demand and memory footprint, then model accuracy is improved, but memory resources increase
Solution Approach 1:
The patent extracts and removes redundant and duplicate samples from the training dataset, keeping only the essential information needed for model training. This reduction in dataset size directly decreases memory footprint while preserving model accuracy, as the remaining samples contain the critical patterns needed for correct classification.
Solution Approach 2:
The patent applies partial action by retaining only the necessary portion of the dataset required for accurate modeling. Instead of using the complete dataset, it selectively keeps samples that provide unique information, thereby reducing memory consumption to the minimum necessary level while maintaining model correctness.
3Measurement precision
If larger datasets are used for training, then model accuracy is improved, but computation time increases
Solution Approach 1:
The patent extracts and removes redundant and duplicate samples from the training dataset, keeping only the essential information needed for model training. This reduction in dataset size directly decreases computation time while preserving model accuracy, as the remaining samples contain the critical patterns needed for correct classification.
Solution Approach 2:
The patent applies partial action by retaining only the necessary portion of the dataset required for accurate modeling. Instead of using the complete dataset, it selectively keeps samples that provide unique information, thereby reducing computation time to the minimum necessary level while maintaining model correctness.
4Adaptability or versatility
If more samples are included in the training set, then model coverage of population variability is improved, but model complexity increases
Solution Approach 1:
The patent extracts and removes redundant and duplicate samples from the training dataset, keeping only the essential information needed for model training. This reduction in dataset size directly decreases computation time while preserving model accuracy, as the remaining samples contain the critical patterns needed for correct classification.
Data Source
AI summary
Methods, apparatus, and machine-readable mediums are described for selecting a training set from a larger data set. Samples are divided into a training set and a validation set. Each set meets one or more conditions. For each class to be modeled, multiple training sets are created. Models are trained on each of the multiple training sets. A size of samples for each class is determined based upon the trained models. A training data set that includes a number of samples based upon the determined size of samples is created.