Rule Set Model Selection Under Accuracy-Interpretability Trade-Off
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
There is a trade-off relation between prediction performance and interpretability in rule set models, making it difficult to find an appropriate model with high performance and interpretability.
Innovation Solution
An information processing apparatus generates and selects a group of rule set models based on prediction performance and rule count constraints, using a greedy algorithm to maximize a region in a coordinate space representing these factors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the number of rules in a rule set model is increased to improve prediction performance, then prediction accuracy improves, but model complexity and interpretability deteriorate
Solution Approach 1:
The patent segments the rule set models into multiple candidate models with different rule counts. The generating unit creates a plurality of rule set models where each model has a different number of rules (first rule count, second rule count, etc.), allowing the system to offer multiple segmented options rather than a single monolithic model. This segmentation enables users to choose models at different complexity levels.
Solution Approach 2:
The patent changes the parameter of rule count to generate different rule set models. By varying the rule count parameter (creating models with different numbers of rules), the system generates a diverse set of models with different prediction performances and interpretability levels. The selecting unit then chooses the optimal model based on this parameter variation.
2Measurement precision
If the number of rules in a rule set model is increased to improve prediction performance, then prediction accuracy improves, but interpretability deteriorates
Solution Approach 1:
The patent segments the rule set models into multiple candidate models with different rule counts. The generating unit creates a plurality of rule set models where each model has a different number of rules (first rule count, second rule count, etc.), allowing the system to offer multiple segmented options rather than a single monolithic model. This segmentation enables users to choose models at different complexity levels.
Solution Approach 2:
The patent changes the parameter of rule count to generate different rule set models. By varying the rule count parameter (creating models with different numbers of rules), the system generates a diverse set of models with different prediction performances and interpretability levels. The selecting unit then chooses the optimal model based on this parameter variation.
3Adaptability or versatility
If multiple rule set models are generated to provide choices, then model selection flexibility improves, but computational resources and processing time increase
Solution Approach 1:
The patent performs preliminary action by pre-generating a plurality of rule set models with different rule counts before the actual model selection process. The generating unit creates these candidate models in advance, so that when the selecting unit needs to choose a model, the work of generating multiple variants has already been completed. This preliminary generation saves time during the selection phase.
Solution Approach 2:
The patent introduces dynamics by making the rule count variable and adjustable. Instead of a fixed rule count, the system dynamically generates models with different rule counts (first rule count, second rule count, etc.), allowing flexible adaptation to different requirements. This dynamic approach enables the system to respond to different selection criteria efficiently.
Data Source
AI summary
An information processing apparatus of the present disclosure includes: a generating unit configured to, based on prediction performance on training data by a rule set model composed of a combination of rules making predetermined prediction on the training data, generate a plurality of rule set models satisfying a constraint rule count representing a constraint on a combinative rule count; and a selecting unit configured to, based on a position corresponding to the rule set model in a space with an axis of prediction performance and an axis of rule count, select and output a model group composed of a combination of the rule set models satisfying a constraint model count representing a constraint on a combinative model count.