Bachelors Thesis - Multiagent Reinforcement Learning with dynamic goals for 5G bandwidth slicing. (Developed custom environments resembling the
5G use case with gym, PyMARL and RLlib frameworks)
- Multi-Student Knowledge Distillation: Developed disentangled DL models with a novel loss function to produce highly uncorrelated feature space, reducing the inference
time significantly by learning only the essential features.
- Developed an indoor localization technique to enhance indoor Li-Fi technology using camera communication.
Paper Github Video