Transaction Anomaly Detection Using Noisy Data Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data transaction anomaly detection systems often generate false alerts due to the preprocessing stage's noise reduction, which can misclassify normal events as anomalous, leading to inefficiencies and reduced accuracy.
Innovation Solution
An augmented anomaly detection model processes noisy data to assign anomaly event indicators, utilizing cluster cells and vector pairs to improve the detection of anomalous transaction events, and incorporates user feedback for model refinement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If noisy data is removed during preprocessing to improve data quality, then measurement precision improves, but loss of information occurs because potentially valid anomalous events are discarded
Solution Approach 1:
The patent converts the harmful noisy data into a beneficial resource by using it to train the anomaly detection model. Instead of discarding noisy data during preprocessing, the system utilizes it to teach the model what constitutes normal variation versus actual anomalies, thereby improving detection accuracy while preserving potentially valid anomalous events
Solution Approach 2:
The system implements feedback by using noisy data to continuously refine and retrain the anomaly detection model. The model learns from noisy data patterns and adjusts its detection thresholds, creating a feedback loop that improves measurement precision without permanently discarding the noisy information
2Speed
If traditional anomaly detection methods are used to identify anomalous events, then detection speed improves, but reliability deteriorates due to false alerts
Solution Approach 1:
The system performs preliminary action by training the anomaly detection model in advance using noisy data before actual anomaly detection begins. This pre-training establishes baseline expectations and detection thresholds, enabling the model to quickly and reliably identify anomalies without generating false alerts during runtime
Solution Approach 2:
The patent applies parameter changes by adjusting the model's detection parameters and thresholds based on patterns learned from noisy data. This dynamic parameter adjustment allows the system to maintain high detection speed while improving reliability by adapting to the specific characteristics of the data being monitored
Data Source
AI summary
Noisy data parsed from raw data can be received. The noisy data indicates first transaction events determined to be noise in the raw data. Using the noisy data, a first detection model can be trained to assign anomaly event indicators to second transaction events. The first detection model can receive an anomaly record. The anomaly record can indicate at least a portion of anomalous transaction events identified in runtime data. The first detection mode can assign the anomaly event indicators to the anomalous transaction events. The anomaly event indicators can indicate levels of severity of the anomalous transaction events identified in the runtime data.