Self-Tuning Object Property Estimator for Streaming Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional machine learning and artificial intelligence systems face challenges in performing training and running algorithms in real-time for streaming objects, particularly due to the procedural and computational complexity of the training phase, which hinders adaptability to changing object characteristics.
Innovation Solution
A self-tuning online estimator technology performs auto-adaptive pattern matching between feature vectors of received objects and object models, allowing for real-time training and estimation of unknown variables through a hierarchical arrangement of estimator stages, with feedback mechanisms for refining object models based on user interaction and actual values.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional machine learning training is performed, then model accuracy is improved, but processing time increases and real-time capability is lost
Solution Approach 1:
The system performs preliminary actions by pre-computing and storing object models with known Y-variable values during offline training. These pre-computed models are then retrieved and matched against incoming streaming objects to produce estimates without requiring real-time training, thus resolving the contradiction between accuracy and training time.
Solution Approach 2:
The patent creates copies of training data in the form of object models that store X-variable to Y-variable mappings. These models are copied and stored in memory structures that enable fast retrieval during online operation, eliminating the need for real-time training computations while maintaining accuracy.
2Adaptability or versatility
If model training is performed online for streaming objects, then adaptability to changing characteristics is improved, but computational complexity increases
Solution Approach 1:
The system dynamically adapts to changing object characteristics by continuously receiving streaming objects with known Y-values and updating the object model database accordingly. This dynamic updating process allows the system to adapt to changing characteristics without requiring complex real-time training algorithms, thus resolving the contradiction between adaptability and computational complexity.
Solution Approach 2:
The patent implements feedback mechanisms where streaming objects with known Y-variable values are used to refine and update the object models. This feedback loop enables continuous improvement of estimation accuracy while using simple database update operations rather than complex computational training, resolving the contradiction between adaptability and complexity.
3Productivity
If simple object models are used, then processing speed is improved, but estimation accuracy deteriorates
Solution Approach 1:
The patent segments the estimation task by dividing object characteristics into X-variables (features) and Y-variables (target properties). It creates separate object models that map X-variables to Y-variables, allowing fast retrieval and matching operations while maintaining comprehensive accuracy through the detailed feature-property relationships stored in the segmented model structure.
Data Source
AI summary
This disclosure describes systems and methods for using an estimator to produce values for dependent variables of streaming objects based on values of independent variables of the objects. The systems and methods may include continuously tuning the estimator based on any objects received with pre-populated values for the dependent variables.