Formula Index for Raw Data Augmentation in Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current machine learning models face challenges in identifying and generating relevant features for effective data augmentation, as existing methods like automated feature engineering, text mining, and knowledge graphs are limited by false positives and constraints on available features, and fail to leverage direct mathematical relationships.
Innovation Solution
The approach involves querying a formula index that stores mathematical formulas to discover new features by mapping identifiers, allowing for the direct extraction of features with mathematical relationships, thereby augmenting the feature set without constraints and reducing false positives.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If automated feature engineering is used to generate features, then the feature set can be expanded, but false positives increase and reliability decreases
Solution Approach 1:
The patent introduces a formula index as an intermediary between existing features and the machine learning model. This formula index contains pre-validated mathematical relationships that serve as a reliable mediator to generate new features without the false positives associated with automated feature engineering. The formula index acts as a trusted knowledge base that bridges the gap between available data and useful features.
2Quantity of substance
If text mining and knowledge graphs are used for data augmentation, then additional features can be discovered, but constraints on available features limit the process
Solution Approach 1:
The formula index serves multiple functions: it validates mathematical relationships, generates new features, verifies feature correctness, and provides a structured approach to feature engineering. This multi-functional approach replaces multiple separate processes (text mining, knowledge graphs, automated feature engineering) with a single universal system that overcomes their individual limitations.
3Quantity of substance
If existing data augmentation techniques are used, then some additional features can be obtained, but the breadth and quality of the feature set remain limited
Solution Approach 1:
The system performs preliminary validation of mathematical relationships by storing verified formulas and their relationships in the formula index before using them for feature generation. This preliminary action ensures that only mathematically sound relationships are used to generate features, guaranteeing their quality and precision before they are applied to the machine learning model.
Data Source
AI summary
Aspects of the invention include techniques for augmenting a feature set with additional raw data. A non-limiting example method includes receiving an input that includes a feature set. The feature set includes a plurality of features. The method includes querying, using the input, a formula index. The formula index includes a plurality of formulas and a plurality of identifiers. One or more identifiers are mapped in the formula index to each respective formula of the plurality of formulas. The method includes, responsive to the querying, returning an output having one or more additional features for the input.