AI Domain Redefinition for Dark Data Utilization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current artificial intelligence systems face challenges in adapting to new scenarios without extensive retraining, struggling with inconsistent and diverse data sets, and are unable to utilize 'dark data' due to predefined domains, leading to inefficiencies in big-data analysis and causality determination.
Innovation Solution
The system dynamically redefines the domain of artificial intelligence functions to include dark data by using an in-memory neural network and ontological models, allowing for the automatic identification and inclusion of previously unused data, thereby enhancing the ability to recognize patterns in new scenarios and improve data analysis efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional AI models are used with predefined domains, then the system operates reliably within known parameters, but it cannot identify or utilize dark data outside its domain, leading to loss of information and inability to adapt to new scenarios
Solution Approach 1:
The patent applies dynamics by making the domain definition dynamic rather than static. The system continuously updates and redefines its domain based on incoming data patterns, allowing it to adapt to new scenarios automatically. This is achieved through mechanisms that detect when data falls outside the current domain and trigger domain expansion, enabling the AI to utilize previously inaccessible dark data without requiring manual retraining.
2Reliability
If humans manually recognize the need for retraining AI models, then retraining can be performed, but this process introduces significant time delays and reduces productivity in analyzing new patterns
Solution Approach 1:
The patent implements feedback mechanisms that automatically monitor data streams for patterns indicating new scenarios. When the system detects data that falls outside its current domain or identifies emerging patterns, it automatically triggers the retraining process without human intervention. This closed-loop feedback system ensures the AI maintains high pattern recognition accuracy while eliminating the time delays associated with manual detection and initiation of retraining.
Solution Approach 2:
The system performs preliminary actions by continuously preparing and pre-processing data in real-time, maintaining trained models ready for immediate deployment. The automatic detection and triggering mechanisms ensure that retraining is initiated at the optimal moment, reducing the lag between identifying new patterns and having a trained model ready for production use.
3Measurement precision
If AI systems require hundreds to thousands of training examples, then the model learns patterns accurately, but this extensive retraining requirement increases time consumption and computational resources
Solution Approach 1:
The patent applies partial action by utilizing only the necessary subset of data required for effective retraining rather than requiring hundreds or thousands of examples. The system intelligently identifies and selects representative training samples from the data stream, performing retraining with minimal but sufficient data. This approach maintains pattern learning accuracy while dramatically reducing the time and computational resources required for retraining.
4Adaptability or versatility
If data from various sources is analyzed, then comprehensive analysis is achieved, but data inconsistency increases complexity and processing difficulty
Solution Approach 1:
The patent applies universality by designing a domain definition and data processing framework that can handle multiple data sources and formats through a unified approach. The system uses universal domain concepts that can encompass diverse data types and sources, applying consistent processing rules regardless of the origin or format of the input data. This universal framework reduces processing complexity while maintaining the ability to analyze comprehensive multi-source data.
Data Source
AI summary
An artificial intelligence implemented method for executing a function by dynamically redefining a domain of the function to include dark data stored in a database. Discrete values are determined from among relevant values of relevant data events. Values that correspond to the discrete values are identified in other data events that are not relevant data events. Each discrete value is linked to each category in which the associated value for any data event corresponds to the discrete value, so as to identify one or more dark categories as categories that are linked to one or more of the discrete values and are not relevant categories. Each dark category is mapped to the function so as to redefine the domain. Data events are instantiated in the in-memory neural network according to the redefined domain. The function is executed in accordance with the redefined domain.