Automated Feature Engineering via Unsupervised Knowledge Graph Embedding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Feature engineering in machine learning is a time-demanding task that requires manual extraction of relevant features using domain knowledge, making it inefficient and labor-intensive, especially when dealing with complex data sets that include unstructured concepts.
Innovation Solution
A method that utilizes unsupervised learning to link input data to an external knowledge graph, expanding columns by associating concepts and generating embedding vectors using a neural network, which can then be used as additional features for predictive analytics tasks without requiring labeled data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual feature engineering is used to extract features using domain knowledge, then the quality and informativeness of features is improved, but the time consumption and labor intensity increase significantly
Solution Approach 1:
The patent replaces manual mechanical feature engineering with an automated neural network system. The neural network automatically learns and extracts features from raw data through unsupervised learning, substituting the manual domain knowledge-based approach with an automated computational system that can process data at scale without proportional increases in time or labor.
Solution Approach 2:
The neural network performs self-learning by automatically identifying patterns and extracting features from data without requiring manual intervention or labeled examples. The system serves itself by autonomously improving its feature extraction capabilities through unsupervised learning mechanisms, eliminating the need for continuous manual feature engineering.
2Reliability
If manual feature engineering is performed to select informative features, then the performance of machine learning algorithms is improved, but the complexity and labor intensity of the process increases
Solution Approach 1:
The patent replaces complex manual feature selection processes with an automated neural network system. The neural network automatically identifies and extracts informative features through unsupervised learning, substituting the complex manual analysis and selection process with a unified automated system that handles both feature extraction and selection simultaneously.
Solution Approach 2:
The neural network performs multiple functions within a single system: it automatically extracts features, selects informative features, and prepares them for machine learning algorithms. This multi-functional approach consolidates what were previously separate manual tasks into one automated process, reducing overall process complexity while maintaining or improving algorithm performance.
3Loss of information
If extensive manual feature engineering is conducted to handle complex unstructured data, then the informativeness of features is improved, but the productivity of the data processing pipeline decreases
Solution Approach 1:
The patent replaces manual feature engineering processes with automated neural network-based feature extraction. The neural network efficiently processes complex unstructured data at scale, automatically identifying and extracting informative features without the productivity losses associated with manual processing. This substitution enables high-volume data processing while maintaining comprehensive information retention.
4Measurement precision
If domain knowledge is manually applied to extract features, then the relevance and discriminating power of features is improved, but the ease of operation and scalability of the process deteriorates
Solution Approach 1:
The patent replaces manual domain knowledge application with automated neural network learning. The neural network automatically learns relevant features and their relationships from data without requiring manual domain knowledge injection, making the process easier to operate and more scalable. The system adapts to different domains and data types through unsupervised learning rather than requiring manual reconfiguration for each application.
Data Source
AI summary
A method, computer system, and computer program product for enhancing feature engineering based on unsupervised learning of associated external knowledge embedding are provided. The embodiment may include receiving, by a processor, input data as a table and a name of a column. The embodiment may also include analyzing the column to identify multisets of concepts or sequences of concepts. The embodiment may further include automatically expanding the column by linking the identified multisets or the sequences of the concepts with corresponding concepts in an external knowledge graph. The embodiment may also include training a neural network to learn embedding vectors of concept multi-sets in the expanded column of the tables, wherein the training is unsupervised without provision of labels of data when the neural network learns an embedding of the multisets of concepts with an objective to minimize a reconstruction error of the identified multisets of concepts.