Prediction Model Retraining With User Feedback for Anomaly Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Prediction models are difficult to develop and often fail to match forecasts due to uncertainties in determining important factors, leading to inaccuracies in anomaly detection.
Innovation Solution
A system for prediction modeling that includes an interface to receive historical data, determines hyperparameters based on a best model, trains using these parameters, detects anomalies, receives user feedback, and retrains the model to improve accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a prediction model is developed using traditional methods, then the model development process is completed, but the model accuracy and alignment with forecasts remain insufficient
Solution Approach 1:
The patent implements feedback loops where user feedback on detected anomalies is incorporated to retrain the prediction model. The system detects anomalies, receives user feedback on those anomalies, and uses this feedback to retrain the model, creating a continuous improvement cycle that enhances prediction accuracy while managing complexity through automated iterative processes
Solution Approach 2:
The patent performs preliminary actions by pre-processing historical data, determining hyperparameters based on initial model training, and establishing baseline prediction models before actual anomaly detection begins. This preliminary setup reduces the complexity of the main prediction task by preparing data and parameters in advance
2Measurement precision
If more factors are included in the prediction model to improve accuracy, then the prediction quality improves, but the difficulty of determining important factors increases
Solution Approach 1:
The system performs self-service by automatically determining hyperparameters based on initial model training and using these hyperparameters to train the final prediction model. The system autonomously identifies important factors through the training process on historical data, reducing the manual difficulty of factor identification while improving forecast alignment
3Reliability
If the model is retrained with user feedback, then the anomaly detection accuracy improves, but the time and computational resources required increase
Solution Approach 1:
The patent implements periodic action by retraining the prediction model at intervals based on accumulated user feedback rather than continuously. The system detects anomalies, collects user feedback over time, and periodically retrains the model with this feedback, balancing improved anomaly detection reliability with reduced time loss compared to continuous retraining
Data Source
AI summary
An interface is configured to receive historical data. A processor is configured to determine a training and a test data set; train models using the training data set to obtain trained models; determine a best trained model of the trained models using the test data set; select hyperparameters associated with the best trained model; generate a prediction model using the hyperparameters and the historical data to obtain a trained prediction model; determine a detected anomaly based on a difference between a forecast and the output of the trained prediction model; provide the forecast, the output of the trained model, and the detected anomaly to an interface; receive user feedback from the interface, wherein the user feedback comprises a false detected anomaly indication indicating that the detected anomaly is not an anomaly; and retrain the trained prediction model using the hyperparameters and the user feedback to obtain a retrained prediction model.