Auxiliary Model Feature Generalization for Predictive Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The challenge is to minimize data collection while maintaining the performance of predictive models, adhering to regulations like GDPR, and reducing user data input requirements without compromising prediction quality.
Innovation Solution
Training an auxiliary model to generalize features, reducing granularity or omitting redundant features, and using dynamic recoding to minimize data collection while maintaining prediction accuracy, with the aid of a decision tree or similar models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If all data is collected and stored indefinitely, then data availability for analysis is improved, but data storage requirements and privacy risks increase
Solution Approach 1:
The patent extracts only the essential features and patterns from the training data that are necessary for model performance, discarding redundant information. This is achieved through feature selection and dimensionality reduction techniques that identify and retain only the most informative attributes while removing unnecessary data elements.
Solution Approach 2:
The patent transforms the data representation by changing parameters such as feature granularity and domain discretization. By adjusting the level of detail in feature representations and optimizing the number of bins or categories, the system achieves better compression ratios while maintaining predictive accuracy within acceptable thresholds.
2Measurement precision
If detailed features are used in predictive models, then prediction accuracy is improved, but data collection burden and processing complexity increase
Solution Approach 1:
The patent segments the feature space into discrete bins or categories, transforming continuous detailed features into manageable discrete segments. This segmentation allows the model to capture essential patterns while reducing the computational burden of processing fine-grained continuous data, effectively balancing accuracy with complexity.
3Reliability
If more features are collected from users, then model performance is improved, but user privacy exposure and data minimization compliance worsen
Solution Approach 1:
The patent extracts only the minimum necessary features required to achieve acceptable model performance, discarding personally identifiable or sensitive information that is not essential for prediction. This selective extraction ensures compliance with data minimization principles while maintaining sufficient model reliability.
Solution Approach 2:
The patent changes the granularity parameter of features to reduce privacy exposure. By coarsening the granularity of sensitive features (e.g., age ranges instead of exact ages, income brackets instead of exact salaries), the system maintains predictive capability while significantly reducing individual identifiability and privacy risks.
Data Source
AI summary
A method, apparatus and a product for data generalization for predictive models. The method comprising: obtaining a training dataset that comprises a plurality of training instances and predicted labels thereof, wherein each training instance is a valuation of a set of features, wherein the set of features comprises a feature having a domain, wherein the predicted label of each training instance is a label predicted thereto by a predictive model; training an auxiliary model using the training dataset; based on the auxiliary model, determining an alternative set of features that is a generalization of the set of features, wherein the alternative set of features comprises a generalized feature having a generalized domain, wherein each value in the generalized domain corresponds to one or more values in the domain; obtaining a generalized instance having a valuation of the alternative set of features; and determining a label for the generalized instance.