Conditional GAN Data Synthesis for New Communication Pattern Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing machine learning systems for processing electronic communications become stale over time, leading to incorrect approvals of new communication types due to lack of training on novel scenarios, resulting in resource loss and security vulnerabilities.
Innovation Solution
Employ a generative adversarial network (GAN) trained to generate synthetic data simulating new communication scenarios, using a conditional tabular GAN (CTGAN) to conditionally generate synthetic training examples based on variable combinations that have not been encountered, enhancing the training of machine learning classifiers to identify risky scenarios proactively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional training techniques are used with known labeled communications, then the machine learning model can be trained initially, but the model becomes stale over time and cannot detect new communication patterns
Solution Approach 1:
The system performs preliminary action by generating synthetic communication data that anticipates future novel communication patterns before they occur in reality. The generative model creates realistic synthetic examples of new communication types, allowing the machine learning model to be proactively retrained on these simulated scenarios, thus preventing staleness and improving adaptability to future threats.
2Adaptability or versatility
If more real communication data is collected for training, then the model coverage increases, but computational resources and time are consumed
Solution Approach 1:
The system creates copies of real communication data by using a generative model to synthesize realistic communication examples that mimic the characteristics of real data. These synthetic copies expand the training dataset coverage without requiring additional real data collection, thus improving model adaptability while avoiding the time and resource costs of gathering more actual communications.
3Reliability
If the machine learning model is continuously retrained with new data, then the model stays updated, but computational resources and processing time are consumed
Solution Approach 1:
Instead of continuously retraining with real new data, the system creates synthetic copies of potential new communication patterns using the generative model. This allows periodic retraining on diverse synthetic data that covers a broad range of possible novel scenarios, maintaining model freshness and reliability while reducing the frequency and resource intensity of retraining operations compared to continuous real-data updates.
Data Source
AI summary
In disclosed techniques a system generates, using a generative model, current synthetic communications, including inputting conditions for the synthetic communications into the trained generative model. The system generates the trained generative model by iteratively performing multiple operations until a discriminator of the generative model determines that synthetic communications output by the generative model satisfy a difference threshold. The operations include: generating, by a generator of the generative model, based on existing communications, a training synthetic communications, determining, by the discriminator of the generative model, differences between the existing communications and the training synthetic communications, and updating the generator based on the differences. Using the current synthetic communications and the existing communications, the system trains another model to evaluate newly initiated communications. The disclosed data synthesis techniques may advantageously enable discovery of concealed patterns, which in turn improves detection of processing systems that execute models trained on the synthetic data.