Multi-tenant ML Platform Ontology Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Identifying appropriate machine learning models, training parameters, and visualizations for novice users can be confusing, and existing systems lack an end-to-end data platform for multiple users to input proprietary data for model training while avoiding overfitting with training data.
Innovation Solution
A method and system that dynamically model multi-tenant data in a machine learning platform by receiving user data, characterizing it, aligning attributes with an ontology, identifying pre-trained models, and recommending them, along with visualizations, to facilitate user interaction and model training within the platform.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a machine learning platform allows multiple users to input proprietary data for model training, then the platform's versatility and adaptability improve, but the complexity of managing and aligning diverse data sets increases
Solution Approach 1:
The patent implements a universal ontology framework that enables the platform to handle diverse data sets from multiple users through a common data model. The ontology serves as a multi-functional interface that can map various data types and structures to standardized concepts, allowing the same platform infrastructure to serve different users and use cases without requiring separate management systems for each data set.
Solution Approach 2:
The patent introduces an intermediary layer consisting of the ontology and recommendation engine that mediates between user-specific data sets and the machine learning model training process. This intermediary automatically aligns diverse data attributes through ontology mapping and recommends appropriate pre-trained models based on data characteristics, thereby reducing the complexity of direct data management while maintaining platform versatility.
2Ease of operation
If the system provides comprehensive model recommendations and visualizations for novice users, then ease of operation improves, but the computational resources and system complexity increase
Solution Approach 1:
The patent implements a self-service recommendation engine that automatically analyzes user-uploaded data sets, characterizes their attributes, and recommends suitable pre-trained models without requiring manual intervention from system administrators or expert users. The system autonomously performs data characterization, ontology alignment, and model recommendation, thereby improving ease of operation for novice users while managing system complexity through automated processes.
Solution Approach 2:
The patent employs pre-trained models that have been previously trained on diverse data sets, allowing the system to provide ready-to-use models immediately upon user data upload. This preliminary action of pre-training models in advance eliminates the need for users to perform complex model training processes, significantly improving ease of operation while the system only needs to manage the curation and recommendation of these pre-trained models.
3Measurement precision
If the system trains models using user-specific training data, then model accuracy for that user improves, but the risk of overfitting increases
Solution Approach 1:
The patent merges user-specific training data with platform-wide pre-trained models through ontology-aligned data integration. By combining the general knowledge captured in pre-trained models with user-specific data characteristics, the system achieves better model accuracy while maintaining generalization capabilities. The ontology framework ensures that user data is integrated in a way that complements rather than overfits to the pre-trained model structures.
Solution Approach 2:
The patent implements a feedback mechanism where the recommendation engine continuously monitors model performance and data characteristics, adjusting model recommendations and training parameters to prevent overfitting. The system uses feedback from data characterization results and model performance metrics to select appropriate pre-trained models and configure training parameters that balance user-specific accuracy with generalization reliability.
Data Source
AI summary
A method, apparatus, system, and computer program code for dynamically modeling multi-tenant data in a machine learning platform. A recommendation engine receives a first data set from a user. The recommendation engine characterizes the first data set to determine data attributes and data characteristics of the first data set. The recommendation engine aligns the data attributes of the first data set with a second data set according to an ontology. Based on the data characteristics of the first data set, the recommendation engine identifies a set of pre-trained models that was trained from training parameters selected from data attributes and data characteristics of a second data set. The recommendation engine recommends the set of pre-trained models to the user.