Decision Tree Pruning for Variable Deprecation in Risk Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Machine learning models with large numbers of variables face accuracy issues and maintenance challenges due to variable deprecation, leading to costly and time-consuming retraining processes when variables become unavailable.
Innovation Solution
Implementing robust neural network and decision tree models that self-compensate for deprecated variables by dropping likely-to-be-deprecated variables during training and adjusting predictions using dropout and pruning techniques, respectively, without the need for retraining or forming new models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional machine learning models are used with large numbers of variables, then prediction coverage is comprehensive, but maintenance complexity increases and retraining is required when variables are deprecated
Solution Approach 1:
The patent applies preliminary action by pre-training the neural network with dropout techniques during the training phase, where certain variables are randomly dropped to simulate future deprecation scenarios. This prepares the model in advance to handle variable deprecation without requiring retraining when variables are actually deprecated, thus reducing maintenance complexity while preserving prediction accuracy.
Solution Approach 2:
The patent uses parameter changes by dynamically adjusting the network architecture through pruning techniques after training. When variables are deprecated, the corresponding neurons and connections are pruned from the network, changing the structural parameters of the model to adapt to the reduced variable set without requiring complete retraining, thereby maintaining reliability while reducing maintenance burden.
2Ease of operation
If variables are deprecated from the model, then maintenance effort is reduced, but prediction accuracy deteriorates
Solution Approach 1:
The patent applies preliminary action by pre-training the neural network with dropout techniques during the training phase, where certain variables are randomly dropped to simulate future deprecation scenarios. This prepares the model in advance to handle variable deprecation without requiring retraining when variables are actually deprecated, thus reducing maintenance complexity while preserving prediction accuracy.
Solution Approach 2:
The patent implements self-service through automated pruning techniques that enable the model to automatically adapt to variable deprecation. The system autonomously identifies and removes deprecated variables from the network architecture without requiring manual intervention or complete retraining, allowing the model to maintain prediction accuracy while reducing maintenance effort through self-adjustment.
3Reliability
If retraining is performed when variables are deprecated, then prediction accuracy is maintained, but time and computational resources are consumed
Solution Approach 1:
The patent applies preliminary action by pre-training the neural network with dropout techniques during the training phase, where certain variables are randomly dropped to simulate future deprecation scenarios. This prepares the model in advance to handle variable deprecation without requiring retraining when variables are actually deprecated, thus reducing maintenance complexity while preserving prediction accuracy.
Solution Approach 2:
The patent uses dynamics by implementing a flexible, adaptive model architecture that can dynamically adjust to variable deprecation through pruning. Rather than static retraining schedules, the system dynamically prunes deprecated variables from the network, allowing continuous adaptation without time-consuming retraining processes while maintaining prediction accuracy.
4Reliability
If dropout techniques are used during training, then robustness to variable deprecation is improved, but training complexity increases
Solution Approach 1:
The patent applies preliminary action by pre-training the neural network with dropout techniques during the training phase, where certain variables are randomly dropped to simulate future deprecation scenarios. This prepares the model in advance to handle variable deprecation without requiring retraining when variables are actually deprecated, thus reducing maintenance complexity while preserving prediction accuracy.
Data Source
AI summary
Various techniques for determining risk assessment predictions and decisions are disclosed. Certain disclosed techniques include the implementation of decision-tree based models in determining predictions of risk for an operation based on an input dataset. The disclosed techniques include pruning decision trees to compensate for deprecation of variables from the input dataset. Decision trees may be pruned at nodes associated with the deprecated variables to inhibit the decision trees from breaking down during operation on an input dataset having deprecated variables.