The course will introduce the requirements, scenarios and architectures for the next-generation mobile networks, together with their challenges and open issues. We will discuss the central role played by the historical data exchanged among the different network entities and how to distribute computing operations across them to enable automatic and energy efficient extraction of context information and network control.
The core focus of the course is the application of Machine Learning (ML) tools to solve identified mobile networking and computing problems. It will be explained what the usage models are and what they imply in terms of stability, convergence and optimality guarantees. For this, fundamentals of Reinforcement Learning and Artificial Neural Networks / Deep Learning will be given. Moreover, Multi-task Learning, Knowledge Transfer Learning, Continual Learning and Federated Learning paradigms for networked systems will be introduced.
Finally, several ML algorithms will be tailored for specific case studies. We will examine the automatic control of base station operation modes to solve the Energy-Quality of Service trade-off; and how to build models for mobile traffic prediction, classification and anomaly detection using real data from mobile operators.
The course covers Supervised, Unsupervised, Semi-Supervised and Reinforcement Learning applications to mobile networking and computing.
The core focus of the course is the application of Machine Learning (ML) tools to solve identified mobile networking and computing problems. It will be explained what the usage models are and what they imply in terms of stability, convergence and optimality guarantees. For this, fundamentals of Reinforcement Learning and Artificial Neural Networks / Deep Learning will be given. Moreover, Multi-task Learning, Knowledge Transfer Learning, Continual Learning and Federated Learning paradigms for networked systems will be introduced.
Finally, several ML algorithms will be tailored for specific case studies. We will examine the automatic control of base station operation modes to solve the Energy-Quality of Service trade-off; and how to build models for mobile traffic prediction, classification and anomaly detection using real data from mobile operators.
The course covers Supervised, Unsupervised, Semi-Supervised and Reinforcement Learning applications to mobile networking and computing.
- Docente: Paolo Dini