ELM Ensemble Retraining with Human Feedback for Stable Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Neural networks face challenges in achieving consistent accuracy during frequent retraining, limiting their use in applications requiring rapid updates.
Innovation Solution
A weighted ensemble of Extreme Learning Machines (ELMs) is used, where each ELM has varying parameters, and a validation test determines weights, allowing for rapid training and retraining based on user feedback.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If neural networks are frequently retrained to adapt to new data, then adaptability improves, but accuracy consistency deteriorates
Solution Approach 1:
The patent segments the neural network into multiple independent ELMs with different parameter configurations. Each ELM is trained on the same data but with varying parameters, creating a diverse ensemble. This segmentation allows the system to maintain accuracy consistency by selecting and weighting the most accurate individual ELMs while retaining adaptability through frequent retraining of the ensemble.
Solution Approach 2:
The patent applies parameter changes by systematically varying ELM parameters (such as hidden layer weights, biases, or activation functions) across different ELMs in the ensemble. This parameter diversity ensures that when new data arrives and retraining occurs, the ensemble maintains performance consistency through the weighted combination of multiple parameter configurations rather than relying on a single fixed model.
2Adaptability or versatility
If traditional neural networks are retrained frequently, then adaptability to new data improves, but training time and computational resources increase
Solution Approach 1:
The patent segments the training process into independent parallel ELM trainings. Each ELM can be trained separately and simultaneously, allowing for efficient utilization of computational resources. When new data arrives, the ensemble can be updated by training individual ELMs in parallel rather than sequentially, significantly reducing the total training time while maintaining adaptability.
Solution Approach 2:
The patent employs multiple simple ELMs that can be quickly trained and discarded if necessary, rather than investing extensive time in training a single complex neural network. Each ELM is relatively simple and fast to train, allowing the system to frequently update the ensemble with new data at low computational cost and time investment.
3Device complexity
If a single neural network is used, then model simplicity is maintained, but accuracy and robustness deteriorate
Solution Approach 1:
The patent merges multiple simple ELMs into an ensemble model where each ELM maintains simplicity individually, but their combination achieves superior accuracy and robustness. The weighted ensemble combines the predictions of multiple ELMs, leveraging their diverse parameter configurations to improve overall performance while keeping each individual component simple and easy to understand.
Data Source
AI summary
An improved neural network is disclosed that supports rapid retraining using human feedback. A weighted ensemble of Extreme Learning Machines (ELMs) is used to implement a model. The ensemble of ELMs may be trained in parallel with a variation in individual parameters gridding a parameter set selected to achieve consistent accurate model results when the model is trained and subsequently retrained when user feedback data become available. An exemplary application is the scoring of resumes.