Autonomous Feature Discovery in AI Models via Graph Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Artificial intelligence models trained on static data become outdated in dynamically changing environments, leading to poor predictions as remembered patterns become irrelevant over time.
Innovation Solution
The method involves updating input features for an AI model by analyzing an information space represented as a topological graph, using optimization techniques like ant colony optimization to detect new features and binning them to reduce redundancy, and then retraining the model with the updated feature set.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the AI model uses static training data and fixed features, then the model structure is simple and easy to implement, but the model becomes outdated and inaccurate in dynamically changing environments
Solution Approach 1:
The patent implements dynamic feature selection by continuously monitoring environment changes and updating the feature set automatically. The system transitions from static to dynamic operation by using change detection mechanisms that trigger feature updates when environmental patterns shift, allowing the AI model to adapt its features in real-time without manual retraining.
Solution Approach 2:
The system incorporates feedback loops that monitor prediction performance and environmental changes. When changes are detected, the feedback mechanism triggers a retraining process using updated training data, ensuring the model continuously improves its accuracy while adapting to new environmental conditions.
2Adaptability or versatility
If the AI model continuously updates features and retrainings, then adaptability to changing environments improves, but computational complexity and processing time increase
Solution Approach 1:
The system performs preliminary actions by pre-processing and storing environmental data in a structured format before changes occur. Training data is organized in advance with feature extraction pipelines pre-configured, so when changes are detected, the system can quickly retrieve and process relevant data without performing complex computations from scratch.
Solution Approach 2:
The patent changes key parameters such as the feature set, learning rate, and model architecture based on environmental conditions. Instead of completely retraining the model, the system adjusts parameters dynamically, which reduces computational complexity while maintaining adaptability to changing environments.
3Reliability
If the AI model uses a large feature set for comprehensive coverage, then prediction accuracy improves, but the model becomes more complex and harder to manage
Solution Approach 1:
The system extracts only the relevant features needed for current predictions by filtering out obsolete or irrelevant features. When environmental changes are detected, the feature selection process identifies and extracts only the necessary features for the new conditions, reducing model complexity while maintaining accuracy.
Solution Approach 2:
The feature set is segmented into multiple categories or groups based on their importance and relevance. The system can selectively activate different feature segments based on environmental conditions, allowing comprehensive coverage when needed while simplifying the model by deactivating irrelevant segments.
Data Source
AI summary
Embodiments are directed to a method of performing autonomous learning for updating input features used for an artificial intelligence model, the method comprising receiving updated data of an information space that includes a graph of nodes having a defined topology, the updated data including historical data of requests to the artificial intelligence model and output results associated with the requests, wherein different categories of input data corresponds to different input nodes of the graph. The method may further comprise updating edge connections between the nodes of the graph by performing path optimizations that each use a set of agents to explore the information space over cycles to reduce a cost function, each connection including a strength value, wherein during each path optimization, path information is shared between the rest of agents at each cycle for determining a next position value for each of the set of agents in the graph.