Machine Learning Model Training via Data Segmentation and Output Merging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing machine learning models, such as capsule networks, face inefficiencies in learning and classification times when dealing with large datasets, leading to prolonged processing times and potential loss of detailed features in data classification.
Innovation Solution
The proposed method involves training multiple vector neural network models with divided input data groups, either by region or class, to generate known feature spectra and utilize these spectra for class determination, reducing data load and improving classification accuracy by integrating individual data from multiple models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a single machine learning model is trained on large datasets, then classification coverage is comprehensive, but learning time and processing time become excessively long
Solution Approach 1:
The patent divides the training data into multiple groups and trains multiple separate machine learning models (first, second, and third models) on different data subsets. This segmentation allows each model to process smaller data portions independently, reducing individual training time and enabling parallel processing, thereby resolving the contradiction between comprehensive classification coverage and excessive learning time.
2Loss of time
If data is divided into multiple groups for training multiple models, then processing time is reduced, but system complexity increases
Solution Approach 1:
The patent combines the outputs of multiple trained models (first, second, and third models) through a merging process that integrates their classification results. This merging mechanism allows the system to leverage the advantages of multiple models while presenting a unified classification output, thereby reducing processing time without proportionally increasing operational complexity.
Solution Approach 2:
The patent introduces an intermediary merging process that coordinates the outputs of multiple models. This intermediary layer synthesizes results from different models and data groups, managing the complexity of multiple models through a structured integration approach that maintains system manageability while achieving faster processing.
3Measurement precision
If multiple machine learning models are trained on divided data, then classification accuracy improves through detailed feature capture, but the determination process becomes more complex
Solution Approach 1:
The patent segments the data into different groups (first, second, and third data groups) with distinct characteristics and trains specialized models on each segment. This segmentation enables each model to capture specific feature patterns detailedly, improving overall classification accuracy through specialized processing of different data aspects.
Solution Approach 2:
The patent creates a multi-functional determination system where multiple models, each trained on specific data segments, collectively handle various classification tasks. The system universally processes different data types through the coordinated operation of specialized models, achieving high accuracy while managing complexity through functional specialization.
Data Source
AI summary
Provided is a learning method including (a) preparing a plurality of pieces of data for learning; (b) dividing the plurality of pieces of data for learning into one or more groups to generate one or more input learning data groups; and (c) training M number of machine learning models, wherein (b) includes (b1) dividing the plurality of pieces of data for input into one or more regions to generate, as one of the input learning data groups, a collection of first type divided input data after division belonging to the same region, or (b2) dividing the plurality of pieces of data for learning belonging to one class into one or more groups to generate, as one of the input learning data groups, a collection of second type divided input data after division.