Inter-Node AIML Model Verification via Configuration Instructions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies lack a method to verify whether an AI/ML model trained by one network node functions as expected when used by another network node, due to issues such as differing software library versions, hardware capabilities, and data environments.
Innovation Solution
A method where a first network node provides a model and configuration information to a second network node, enabling the second node to verify the model's performance, data inputs, and output delivery, ensuring the model is correctly set up and functions as intended before use.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If an AI/ML model is trained by one network node and deployed to another network node, then model sharing and collaboration are enabled, but the model may not function as expected due to differing software library versions, hardware capabilities, and data environments
Solution Approach 1:
The patent applies preliminary action by implementing verification procedures before model deployment. The receiving network node performs verification of the model's input data, output data, and processing logic before actually deploying the model, ensuring that the model will function as expected in the target environment. This prevents deployment failures caused by environmental differences.
Solution Approach 2:
The patent implements feedback mechanisms where the receiving network node provides feedback information about the verification results to the training network node. This feedback loop allows the training node to understand whether the model deployed successfully and why, enabling continuous improvement of model deployment processes and handling of environmental variations.
2Reliability
If verification procedures are implemented for AI/ML models before deployment, then model reliability and performance consistency are improved, but the deployment process complexity increases
Solution Approach 1:
The patent segments the verification process into distinct, manageable components: verification of input data (including data types, formats, and quality), verification of output data (including data types, formats, and quality), and verification of processing logic (including algorithms and computational graphs). This segmentation makes the complex verification process more manageable and systematic.
Solution Approach 2:
The patent creates a universal verification framework that can handle multiple verification aspects through a single system. The verification module is designed to be multi-functional, capable of verifying input data, output data, and processing logic simultaneously, thereby reducing overall system complexity despite the comprehensive verification coverage.
3Manufacturing precision
If comprehensive verification of input data, output data, and processing logic is performed, then model correctness is ensured, but the time required for deployment increases
Solution Approach 1:
The patent applies partial action by implementing verification procedures that focus on the most critical aspects of model correctness. Rather than verifying every possible detail, the system performs verification on key elements such as data types, data formats, data quality, algorithms, and computational graphs, which are the most impactful factors for model correctness. This selective approach ensures accuracy while reducing verification time.
Data Source
AI summary
Systems and methods for inter-node verification of models are disclosed. Some embodiments of the present disclosure propose a method for a first network node to provide a model to a second network node together with configurations/instructions/semantics information for verifying (e.g., testing and/or validating) the model. In some embodiments, the model is an Artificial Intelligence (AI) and/or Machine Learning (ML) model. This may enable a first network node, responsible for training an AIML model and providing it to a second network node, to specify whether, when, and how the model can, should, or must be verified by the second network node prior to using the model. This allows the first network node to ensure that the model, for which it is responsible, is correctly set up, applied, or installed by the second network node and works as expected before the second network uses the model, for instance, for inference.