Deep Generative Modeling for Communication Systems Testing and Data Sharing
Abstract
A common problem that limits distribution of real-world RF data is data sensitivity. Data often contain proprietary information or, in the case of military applications, the data may actually be classified. The goal of the proposed research is to investigate generative models and maps that will remove or obfuscate sensitive information from a population while otherwise being faithful to the target distribution. In addition to seeking to solve the population obfuscation problem, several potential approaches will be presented in this poster as well as initial experiments for investigating and testing these approaches.
Deep Generative Modeling for Communication Systems Testing and Data Sharing
Volstorff A
A common problem that limits distribution of real-world RF data is data sensitivity. Data often contain proprietary information or, in the case of military applications, the data may actually be classified. The goal of the proposed research is to investigate generative models and maps that will remove or obfuscate sensitive information from a population while otherwise being faithful to the target distribution. In addition to seeking to solve the population obfuscation problem, several potential approaches will be presented in this poster as well as initial experiments for investigating and testing these approaches.