This research project, conducted from July to December 2024, focused on synthesizing network traffic across diverse online platforms (domains) while preserving content. The goal was to enhance content moderation and fake news detection. I developed and implemented a novel Content-Informed, Domain-Informed Variational Autoencoder (CI-DI-VAE) to disentangle content and platform features in network traffic. The synthesized traces were evaluated through downstream classification, demonstrating improved cross-platform generalization.
Cross-Domain Network Traffic Synthesis
July 2024
research machine-learning deep-learning
Led research on synthesizing network traffic across diverse online platforms while preserving content.