AI Model Training for DBMS Status Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing AI models for predicting the status of database management systems (DBMS) face challenges due to training inequity and runtime deviation, leading to biased and unreliable predictions, which can be exacerbated by varying knowledge levels among experts and changes in workload.
Innovation Solution
A computer-implemented system and method that generates ingested data by randomly changing feature records in testing data to compute inequity ratios, compares runtime data with ingested data to identify matching records, and adjusts feature weights to retrain the AI model, ensuring robustness and trustworthiness by maintaining inequity and runtime deviation below thresholds.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If AI models are trained using existing training data without intervention, then the model training process is simple and fast, but the model produces biased and unreliable predictions due to training inequity
Solution Approach 1:
The system performs preliminary actions by generating ingested data through random changes to testing data before the actual training process. This pre-processing step creates modified training data that addresses inequity issues, allowing the model to learn from balanced data distributions and produce reliable predictions without requiring complex interventions during training.
Solution Approach 2:
The system implements feedback mechanisms by comparing runtime data with ingested data to identify matching records, then using this feedback to adjust feature weights and retrain the model. This iterative feedback loop continuously improves prediction reliability by correcting biases identified during runtime operation.
2Reliability
If the system generates ingested data by randomly changing feature records to compute inequity ratios, then training inequity is mitigated, but the data processing time and computational complexity increase
Solution Approach 1:
The system applies partial action by randomly changing only a subset of feature records in the testing data to generate ingested data. This selective modification approach mitigates training inequity effectively while minimizing the computational overhead and data processing time compared to transforming the entire dataset.
Solution Approach 2:
The system changes parameters of the training data by generating ingested data with modified feature values. These parameter changes create balanced data distributions that address inequity issues, allowing the model to learn from transformed data without requiring extensive processing time.
3Reliability
If the system compares runtime data with ingested data to identify matching records, then runtime deviation is detected, but the computational overhead and processing complexity increase
Solution Approach 1:
The system creates a copy of the training data structure by generating ingested data that mirrors the original testing data format. This copying approach enables efficient comparison between runtime data and ingested data to identify matching records, detecting runtime deviation accurately without introducing excessive computational overhead.
Solution Approach 2:
The system extracts key information by comparing runtime data with ingested data to identify matching records, then uses this extracted information to adjust feature weights. This extraction approach focuses computational resources on the most critical deviations, improving detection accuracy while minimizing overall processing complexity.
4Reliability
If the system adjusts feature weights based on inequity ratios and matching records, then the AI model becomes more trustworthy and explainable, but the model retraining process becomes more complex
Solution Approach 1:
The system applies local quality by adjusting feature weights selectively based on inequity ratios and matching records identified during comparison. This targeted adjustment approach improves model trustworthiness and explainability by addressing specific biased features without requiring complete model retraining, thus reducing overall complexity.
Solution Approach 2:
The system implements dynamics by making feature weights adjustable and adaptable based on runtime conditions and identified deviations. This dynamic weight adjustment allows the model to evolve and improve its trustworthiness continuously without requiring complex retraining procedures, as the weights can be modified directly based on current data conditions.
Data Source
AI summary
One or more systems, devices, computer program products and/or computer-implemented methods of use provided herein relate to training an AI model to predict status of a DBMS. The computer-implemented system can comprise a memory that can store computer executable components. The computer-implemented system can further comprise a processor that can execute the computer executable components stored in the memory, wherein the computer executable components can comprise a data ingestion component that can use testing data of an AI model to generate ingested data by randomly changing one or more records of at least one feature comprised in the testing data, wherein the ingested data can be used to compute a first ratio indicative of inequity of the at least one feature. The computer executable components can further comprise a training component that can train the AI model using at least the first ratio to predict a status of system.