Game Theoretic Deep Neural Networks for ML Model Generation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The generation of machine learning models is complex and time-consuming, especially when multiple models are required to achieve desired outcomes, and there is a need for secure control over datasets used in model generation to ensure compliance with privacy regulations.
Innovation Solution
A machine learning model generation platform that enables the creation of portable and scalable AI workflows, allowing for end-to-end orchestration, easy experimentation, and efficient packaging of AI pipelines, along with secure data usage through a representative dataset and secure environment for model training and validation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional machine learning model generation methods are used, then model training can be performed, but the process becomes complex and time-consuming when multiple models are required
Solution Approach 1:
The patent segments the model generation process into distinct components: a generator network that creates candidate models and a critic network that evaluates them. This segmentation allows parallel processing and independent optimization of each component, reducing overall complexity while improving generation efficiency through specialized functional modules.
Solution Approach 2:
The system performs preliminary actions by pre-training the generator and critic networks before actual model generation. The generator learns to create plausible model configurations, and the critic learns to evaluate them, so that during deployment, the system can rapidly generate and assess multiple model candidates without starting from scratch each time.
2Productivity
If datasets are shared for model generation, then model training can proceed, but secure control over data access and usage is compromised
Solution Approach 1:
The patent introduces an intermediary secure environment that acts as a mediator between the dataset owner and the model generation process. The representative dataset is processed within this controlled intermediary space, allowing model training to proceed while maintaining data security through enforced access controls and usage policies that prevent direct access to sensitive data.
Solution Approach 2:
Instead of sharing the original sensitive dataset, the system creates and uses a representative dataset that copies only the essential statistical properties and patterns needed for model training. This copying approach enables model development while preserving the security and privacy of the source data, as the representative dataset can be safely shared without exposing sensitive information.
Data Source
AI summary
A method includes using a generator to generate a first result, providing the first result to a critic, generating a first surprise factor based on providing the first result to the critic, based on the first surprise factor, using the generator to generate a second result, providing the second result to the critic, generating a second surprise factor based on providing the second result to the critic, based on the second surprise factor, determining that the generator has generated a most surprising result, and presenting the most surprising result in a graphical user interface.