No-Code ML Toolkit for Automated Model Training
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Business developers lack the technical expertise and time to create and manage machine learning models, as building these models requires advanced knowledge in computing and data exploration, making it a complex process.
Innovation Solution
A metadata-driven toolkit for building and executing machine learning models in a no-code software environment, allowing users to create and run models without coding expertise by defining model intentions and requirements through drop-down menus and input fields, which automatically translates these into metadata for model building and training.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If traditional machine learning model creation processes are used, then model accuracy and performance can be achieved, but the complexity of the process and the need for advanced technical expertise increase significantly
Solution Approach 1:
The patent introduces an automated model generation system that acts as an intermediary between business developers and machine learning models. This system automatically generates, trains, and deploys ML models based on simple business rules and requirements defined by non-technical users, eliminating the need for business developers to directly engage in complex model creation processes while maintaining model accuracy.
Solution Approach 2:
The system enables business developers to self-serve by providing them with tools to define model requirements through simple interfaces (such as YAML configurations or UI forms). The automated system then handles the complex tasks of model generation, training, and optimization independently, allowing business developers to obtain accurate ML models without needing advanced technical expertise.
2Reliability
If traditional machine learning model creation processes are used, then functional models can be built, but the time required for model development and deployment increases
Solution Approach 1:
The system performs preliminary actions by pre-defining model templates, data processing pipelines, and training configurations. When a business developer initiates model creation, the system leverages these pre-prepared components to rapidly generate and train models, significantly reducing development time while ensuring functional reliability through proven templates and configurations.
Solution Approach 2:
The system allows business developers to specify high-level parameters (such as model type, data sources, and performance targets) without needing to configure complex technical parameters. The automated system then handles parameter optimization, hyperparameter tuning, and model selection, reducing the time required for model development while maintaining functionality through automated parameter adjustment.
3Adaptability or versatility
If business developers attempt to create machine learning models independently, then model customization for business needs improves, but the technical expertise required increases beyond their capabilities
Solution Approach 1:
The patent segments the model creation process into distinct, manageable components that business developers can interact with through simple interfaces. Users can define business requirements, select from predefined model types, specify data sources, and set performance targets without needing to understand the underlying complexity. The system handles each segment automatically, maintaining ease of operation while enabling customization.
Solution Approach 2:
The system provides universal tools and templates that can be applied across different business scenarios and model types. Business developers can use the same simplified interface and configuration approaches for various ML tasks (classification, regression, prediction), enabling model customization for specific business needs without requiring different technical expertise for each model type.
Data Source
AI summary
Various embodiments of the present technology include systems and methods for building, training, and executing new machine learning models via a no-code user environment. In some embodiments, a model definition is created by a user via a no-code machine learning model development toolkit and the model is queued for creation. A machine learning engine then implements processes described herein to build a machine learning model based on the model definition, train the model, automatically clean associated data for input into the model, and create a model instance to be stored in a model datastore. When new records are created or received, they may trigger the machine learning engine to run the model against the new record and provide valuable output.