Dynamic Dataset Feedback for Up-to-Date ML Inference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current machine learning models are static and require retraining with new data when knowledge advances, leading to inefficiencies in cross-organizational workstreams due to siloed knowledge management systems that protect proprietary information.
Innovation Solution
Dynamic refinement and inference of datasets using machine learning models, allowing for the creation of user-, group-, or organization-specific datasets that interact with models through a segmental feedback process, enabling precise, up-to-date, and tailored responses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If static datasets are used to train machine learning models, then the models can be trained initially, but they become outdated when knowledge advances or datasets change, requiring retraining
Solution Approach 1:
The patent implements dynamic datasets that automatically update and refine themselves over time without requiring full retraining of the machine learning model. The dataset evolves continuously by incorporating new information and correcting errors, allowing the model to maintain accuracy while avoiding the time-consuming retraining process.
Solution Approach 2:
The system incorporates feedback mechanisms where the machine learning model's outputs are evaluated against ground truth data, and correction signals are fed back to refine the dynamic dataset. This continuous feedback loop ensures the dataset remains accurate and up-to-date, maintaining model reliability without repeated retraining.
2Reliability
If knowledge management systems are siloed to protect proprietary information, then security is maintained, but cross-organizational workstreams become inefficient
Solution Approach 1:
The patent introduces dynamic datasets as an intermediary layer between siloed knowledge management systems. These datasets can be selectively shared across organizations with controlled access, allowing cross-organizational collaboration while maintaining security boundaries. The intermediary enables efficient workstreams without compromising proprietary information protection.
Data Source
AI summary
Computer-implemented systems and methods for dynamic refinement and inference of datasets are disclosed herein. The systems and methods may include a segmental feedback process to enhance the context, analysis, or reasoning of data in datasets used in machine learning models to provide refinement, enrichment, and inference by using the segmental feedback process to create a dynamic dataset (DD) to inform machine learning models, such as generative AI models or LLMs. The DD may be created based on a user, group of users, organization, or subject matter. A notification indicating when a user’s decision, action, or choice deviates from the machine learning model’s expected outcome provides an opportunity for the user to input feedback or the systems and methods to capture context to further refine, enrich, and enhance the DD. The systems and methods may apply the inferences to workstreams, analysis, or decisions for a user, group of users, or organization.