- Developed a creative data-driven model for synthesizing time-series datasets to boost the accuracy and fidelity of downstream tasks.
- Achieved State-of-The-Art results on 3 network security datasets with an accuracy of 81.1% and beat the DoppelGANger paper.
- The synthesized dataset improved the accuracy of the tested models by 5%.
- Trained and tested the model on a dataset of 100,000 data points on distributed GPUs.
yelnady/Data-Synthesizer-for-Improving-Network-Security-Models
A creative data-driven model for synthesizing time-series datasets to boost the accuracy and fidelity of downstream tasks.
Jupyter Notebook