Machine Learning Ensemble Segmentation for Accuracy and Training Time
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current machine learning (ML) techniques face challenges in constructing optimal ensembles and efficiently training multiple algorithms, leading to computational burdens and suboptimal performance in classification tasks.
Innovation Solution
The proposed solution involves using information theoretic methods to reduce training data requirements, prioritize classification accuracy, and fuse multiple ML techniques through a three-layer process, including synthetic data generation and analytical hierarchical processing to optimize and rank ML methods based on their biases and performance metrics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple machine learning algorithms are combined in an ensemble to improve classification accuracy, then the accuracy is improved, but the computational burden and training time increase significantly
Solution Approach 1:
The patent segments the ensemble learning process into distinct layers: a first layer of machine learning algorithms processes the original data, and a second layer of algorithms processes the outputs from the first layer. This segmentation allows for more efficient training by organizing algorithms hierarchically rather than training all algorithms simultaneously on the complete dataset, thus reducing overall training time while maintaining classification accuracy.
2Measurement precision
If multiple machine learning algorithms are combined in an ensemble to improve classification accuracy, then the accuracy is improved, but the device complexity increases
Solution Approach 1:
The patent divides the complex ensemble into manageable layers with clear functional distinctions. The first layer handles initial classification tasks, while the second layer refines results by processing first-layer outputs. This segmentation reduces construction complexity by providing a structured framework for organizing multiple algorithms, making the ensemble more manageable and easier to implement compared to traditional flat ensemble structures.
3Productivity
If synthetic data is generated to reduce training data requirements, then the training efficiency is improved, but the data processing complexity increases
Solution Approach 1:
The patent generates synthetic training data in advance before the main training process. By pre-generating synthetic data that mimics the characteristics of real data, the system reduces the amount of actual training data needed and accelerates the training process. This preliminary action of data generation separates the complex data creation step from the training step, improving overall training efficiency while managing complexity through staged processing.
Data Source
AI summary
A method can include determining a cell of a grid to which a first feature and a second feature of each of a plurality of input/output examples maps, determining an average of respective features of the cell to generate respective level 2 synthetic feature vectors, for each cell with an input/output example of the input/output examples mapped thereto, generating a sub-grid of cells and map the input/output examples mapped to a cell of the sub-grid, determining an average of respective features to generate respective level 1 synthetic feature vectors comprising the average of the respective features, training the ML technique using the level 2 synthetic feature vector, testing the trained ML technique using the level 1 synthetic feature vector of each sub-cell, and further testing the trained ML technique using the input/output examples to generate a class and confidence for each of the input/output examples.