Multi-layer Classifier Fusion for Prediction Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Fusing the outputs of different types of classifiers in machine learning is challenging, as existing methods struggle to effectively combine predictions from various classifier types to improve prediction accuracy.
Innovation Solution
A method is introduced where multiple classifiers of different types are trained in layers, with each layer generating new training data by combining predictions and original data, and subsequent layers build upon these fused data sets to enhance prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple classifiers of different types are trained and their outputs are fused, then prediction accuracy is improved, but the complexity of the system increases
Solution Approach 1:
The system segments the classification task into multiple layers, where each layer contains classifiers of different types (e.g., SVM, Random Forest, Neural Networks). Each layer processes predictions independently and passes results to the next layer, breaking down the complex fusion problem into manageable segments that can be trained and optimized separately.
Solution Approach 2:
The patent introduces a hierarchical dimension to the classifier fusion problem by organizing classifiers into multiple levels. Instead of attempting to fuse all classifiers simultaneously in a single complex model, the system adds a vertical dimension where predictions flow through successive layers, each contributing to the final decision in a structured manner.
2Loss of information
If classifiers are trained on combined data with predictions, then new information is generated, but the training process becomes more time-consuming
Solution Approach 1:
Each layer of classifiers is trained in advance on datasets that include both original features and predictions from previous layers. This preliminary training allows the system to pre-compute optimal decision boundaries and fusion strategies, so that when actual predictions are needed, the system can quickly apply the pre-trained models without performing complex real-time optimization.
Solution Approach 2:
The training process maintains continuity by using the predictions from one layer as additional features for training the next layer. This creates a continuous flow of information where each layer builds upon the previous layer's output, allowing the system to progressively refine predictions without restarting the training process from scratch at each stage.
Data Source
AI summary
A method of generating a predictor to classify data includes: training each of a plurality of first classifiers arranged in a first level on current training data; operating each classifier of the first level on the training data to generate a plurality of predictions; combining the current training data with the predictions to generated new training data; and training each of a plurality of second classifiers arranged in a second level on the new training data. The first classifiers are classifiers of different classifier types, respectively and the second classifiers are classifiers of the different classifier types, respectively.