Security in Machine Learning Models

Project ID
Machine learning has been playing a very important role these days, however, the security is also concerned to researchers. I would like to run deep learning models for several tasks and check whether there is any information leakage out of these models.
Use of FutureSystems
I hope to run tensorflow with the help of GPU and CUDA. So ROMEO is more preferable.
Scale of Use
I will need about 2 months to do this research.