Ensemble ML Classifiers for Streaming Data Concept Drift
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional machine learning models struggle with real-time data processing due to time-intensive retraining, concept drift, and catastrophic forgetting, particularly when dealing with high-volume streaming data.
Innovation Solution
A dynamically updated ensemble of machine learning models that continuously learns from unlabeled data streams, replacing older models with newly trained ones while archiving them for reintroduction, thereby adapting to concept drift and retaining previous knowledge.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If ML models are retrained periodically on static historical data, then model accuracy is maintained, but real-time analytics capability is lost due to time-intensive retraining
Solution Approach 1:
The patent segments the continuous data stream into discrete batches and processes them incrementally through the ensemble of ML models. Each model in the ensemble is trained on specific time windows of the data stream, allowing the system to handle real-time data without requiring complete retraining on the entire dataset. This segmentation enables both real-time processing and maintained accuracy.
Solution Approach 2:
The patent implements a dynamically updated ensemble where ML models are continuously added and removed based on the age and performance of the data they were trained on. The ensemble composition changes over time to adapt to concept drift while maintaining computational efficiency. This dynamic structure allows the system to process real-time data streams without the time penalty of periodic full retraining.
2Adaptability or versatility
If ML models are retrained on newer data windows, then concept drift is addressed, but retraining time increases causing learnings to lag behind data stream changes
Solution Approach 1:
Instead of fully retraining all models on every new data window, the patent applies partial action by selectively updating only certain models in the ensemble based on performance metrics and data characteristics. This partial updating approach reduces retraining time while still addressing concept drift effectively, preventing learnings from lagging behind data stream changes.
Solution Approach 2:
The patent changes the parameter of model training frequency and scope dynamically. Rather than uniform periodic retraining, the system adjusts which models are retrained and how often based on detected concept drift patterns and model performance degradation. This parameter adaptation reduces unnecessary retraining time while maintaining adaptability to data distribution changes.
3Productivity
If ML models are updated dynamically on streaming data, then real-time processing is achieved, but catastrophic forgetting of previously learned knowledge occurs
Solution Approach 1:
The patent merges multiple ML models into an ensemble structure where each model retains knowledge from its training period. By combining predictions from multiple models rather than using a single continuously retrained model, the system preserves previously learned knowledge while adapting to new data patterns. This merging approach enables real-time processing without catastrophic forgetting.
Solution Approach 2:
The patent implements a lifecycle management system for ML models where older models are gradually discarded from the ensemble as newer models are added, but archived models can be recovered and reintroduced when their knowledge becomes valuable again. This selective discarding and recovering mechanism allows the system to maintain real-time processing capability while preventing complete loss of previously learned knowledge patterns.
Data Source
AI summary
Aspects of the present disclosure provide systems, methods, and computer-readable storage media that support dynamically updated ensemble-based machine learning (ML) classification. An ensemble of ML classifiers may be created from a plurality of trained ML classifiers. These initial ML classifiers may be trained using labeled data to generate predictions based on input data. When an unlabeled data stream is received, the unlabeled data stream may be provided as input to the ensemble to generate predictions. After obtaining labels for the received data, the labels and the unlabeled data stream may be used to train new ML classifiers. The new ML classifiers may replace older ML classifiers in the ensemble. In this manner, the ensemble of ML classifiers is used to perform predictions on high volume streaming data while being dynamically updated with ML classifiers that have learned changes in statistical distribution across more recent input data.