Education

• 

  Universität stuttgart

  2017 - 2020

  Master of science - ms electrical, electronics and communications engineering

  Research Thesis: Autoencoder-Based Communication Systems with Deep Reinforcement LearningIn this research thesis I used deep RL to enable training of communication systems with an unknown channel model. It was shown that with this approach is possible to build, train and run a communication system OTA and reach the same performance as QPSK (benchmark).Master Thesis: Optimizing the Learning-Behavior of an Artificial Intelligence for Nonlinear Optical CommunicationsIn this work I… Show more Research Thesis: Autoencoder-Based Communication Systems with Deep Reinforcement LearningIn this research thesis I used deep RL to enable training of communication systems with an unknown channel model. It was shown that with this approach is possible to build, train and run a communication system OTA and reach the same performance as QPSK (benchmark).Master Thesis: Optimizing the Learning-Behavior of an Artificial Intelligence for Nonlinear Optical CommunicationsIn this work I first dealt with hyper-parameters optimization for an autoencoder model for coherent and nonlinear optical communication using Bayesian optimization. The result was improved in about 25% when comparing to a basic grid search to tune the hyper-parameters. Later, the supervised manner was compared with a RL approach. The final result proved that the model could be trained without any knowledge of the channel distribution besides the Kerr-nonlinearity and chromatic dispersion in the fiber. Show less

• 

  Universidade estadual de campinas

  2009 - 2014

  Bachelor of science - bs computer engineering

  Bachelor Thesis: Design and Implementation of Turing's Unorganized Network