Semi-Supervised Classification Model Training Using Expectation Maximization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Supervised learning algorithms require extensive human labor and resources for data labeling, making it complex and expensive to fully label entire training datasets, which hinders efficient machine learning model training.
Innovation Solution
A classification model training method using a probability-based semi-supervised learning approach with expectation maximization, initializing weight values, mean vectors, and covariance matrices to iteratively update and converge these parameters for accurate classification model training.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If supervised learning algorithms are used to train classification models, then prediction accuracy is improved, but extensive human labor and resources are required for data labeling
Solution Approach 1:
The patent applies partial action by requiring only a subset of training data to be labeled rather than the entire dataset. The semi-supervised learning system uses a small portion of labeled data combined with a larger portion of unlabeled data, reducing the labeling burden while maintaining effective model training through iterative algorithms that leverage both labeled and unlabeled information.
2Measurement precision
If supervised learning algorithms are used to train classification models, then prediction accuracy is improved, but human labor and costs increase
Solution Approach 1:
The system uses partial action by processing only a subset of data with full labeling while handling the remainder with semi-supervised learning. This reduces the quantity of human labor required for labeling while still achieving accurate predictions through the combination of labeled subset and unlabeled majority, processed via iterative algorithms.
Solution Approach 2:
The patent introduces an intermediary semi-supervised learning algorithm that bridges the gap between fully supervised and unsupervised learning. This intermediary system uses the limited labeled data as a guide to progressively label and utilize the unlabeled data, reducing direct human labeling effort while maintaining prediction accuracy.
3Measurement precision
If fully labeled training datasets are used, then model training accuracy is improved, but training time and resources increase
Solution Approach 1:
The system applies partial action by using only a subset of data for full labeling while processing the remaining data through semi-supervised learning. This reduces the time and resources required for complete labeling while maintaining model training accuracy through iterative algorithms that effectively utilize both labeled and unlabeled portions of the dataset.
Data Source
AI summary
A computer trains a classification model. (A) An estimation vector is computed for each observation vector using a weight value, a mean vector, and a covariance matrix. The estimation vector includes a probability value for each class of a plurality of classes for each observation vector that indicates a likelihood that each observation vector is associated with each class. A subset of the plurality of observation vectors has a predefined class assignment. (B) The weight value is updated using the computed estimation vector. (C) The mean vector for each class is updated using the computed estimation vector. (D) The covariance matrix for each class is updated using the computed estimation vector. (E) A convergence parameter value is computed. (F) A classification model is trained by repeating (A) to (E) until the computed convergence parameter value indicates the mean vector for each class of the plurality of classes is converged.