Tree-Based Learner Automated Reconfiguration via Sub-Model Addition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional tree-based learners require manual intervention and expertise to adapt to changes in ground-truth data, leading to potential model bias and reduced performance when data sets become obsolete or feature changes occur during runtime.
Innovation Solution
A method and apparatus for managing a tree-based learner that includes an initialisation phase and an automated reconfiguration phase, where the tree-based learner is trained with a first set of specimen data and then reconfigured by adding a new sub-model to process new feature values in a new set of specimen data, using a comparison to determine necessary changes and selecting or training sub-models for adaptation without manual intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual intervention and expert investigation are used to adapt tree-based learners to changes in ground-truth data, then model accuracy can be maintained, but the complexity of operation and time required for adaptation increases significantly
Solution Approach 1:
The system performs self-adaptation by automatically detecting changes in ground-truth data, comparing new data with historical data, and reconfiguring the model structure without requiring manual expert intervention. The automated reconfiguration phase compares new specimen data with first set of specimen data and automatically adds new sub-models when changes are detected, eliminating the need for manual investigation and expert knowledge.
Solution Approach 2:
The system performs preliminary comparison of new specimen data with historical data before making any model changes. This preliminary action identifies changes in ground-truth data characteristics in advance, allowing the system to proactively reconfigure the model structure before these changes affect model accuracy, rather than reacting after manual investigation is required.
2Reliability
If manual investigation and multiple attempts are made to modify tree structure, then appropriate model updates can be achieved, but the time required for adaptation increases
Solution Approach 1:
The system implements an automated feedback loop where new specimen data is continuously compared with historical data, and model reconfiguration is triggered automatically based on detected changes. This feedback mechanism eliminates the need for multiple manual attempts by providing continuous automated monitoring and immediate automated response when changes are detected, significantly reducing adaptation time.
Solution Approach 2:
The system performs preliminary automated comparison of new data with historical data to identify changes before they impact model performance. This preliminary detection action allows the system to proactively reconfigure the model structure in advance, eliminating the need for time-consuming manual investigation and multiple trial modifications.
3Productivity
If the model is trained on fixed data sets, then initial training can be performed efficiently, but the model becomes biased when data sets become obsolete or features change during runtime
Solution Approach 1:
The system transitions from a static model trained on fixed data sets to a dynamic model that automatically adapts to changing data characteristics. The automated reconfiguration phase continuously monitors for changes in ground-truth data and dynamically updates the model structure by adding new sub-models, allowing the model to maintain accuracy despite data obsolescence or feature changes over time.
Solution Approach 2:
The system performs preliminary comparison of new specimen data with historical training data to detect changes before they cause model bias. This preliminary detection allows the system to proactively reconfigure the model structure in advance, preventing bias from occurring rather than correcting it after it has developed, thus maintaining both training efficiency and adaptability.
Data Source
AI summary
A method is proposed for managing a tree-based learner as following. In an initialisation phase: training the tree-based learner with a first set of specimen data to process input data with an active model chain comprising at least a root model to generate an output, each member of the first set of specimen data having a value of each of a first set of features; in an automated reconfiguration phase: receiving a new set of specimen data; performing a comparison between the new set of specimen data and the first set of specimen data; responding to the executed comparison indicating new values of a feature in the new set of specimen data, by reconfiguring the tree-based learner by the addition of a new sub-model to the active model chain, the new sub-model being trained to process at least the new values of the feature.