The Cost of Learning Dilemma
As AI becomes central to adaptive technology, training and updating models imposes a real cost in communication bandwidth, computation, and network resources — competing with the traffic…
Contact: Federico Mason
For students
Thesis Proposals
Master's thesis topics with the AMCO group, spanning reinforcement learning, control, trustworthy AI, and applied computer vision.
We offer thesis topics across our research areas — reinforcement learning, control systems, algorithmic fairness, anomaly detection, and Industry 5.0 — as well as applied computer vision, including visual anomaly detection, object recognition (YOLO), weakly supervised learning, and active learning. Reinforcement-learning theses can be carried out in collaboration with a company or entirely at the university; in both cases a PhD student is assigned as a mentor for guidance and support. Students are also welcome to propose their own topic of interest — self-proposed topics are discussed with the professors to confirm feasibility and fit with the group's research directions.
Open proposals 10
Edge Computing-Based Automated Driving in Duckietown
Duckietown is a small-scale automated-driving platform where cars navigate a miniature town with pedestrians, traffic signs, and other vehicles, running computer vision and decision-making…
Contact: Federico Chiariotti
Security in Goal-Oriented Communications
Goal-oriented communication lets sensors transmit only the data relevant to the receiver, omitting redundant information — but this efficiency creates a new security risk: an eavesdropper…
Contact: Federico Chiariotti
Deep Reinforcement Learning for Sensor Scheduling
Wake-up radio lets IoT nodes respond selectively to ID- or content-based requests, creating scheduling opportunities when sensor readings are correlated: careful request design can save…
Contact: Federico Chiariotti
Goal-Oriented Integrated Sensing and Communication
Integrated Sensing and Communication (ISAC) uses wireless signals themselves to sense the environment — inferring position, identity, and activity from how people perturb a multi-antenna…
Contact: Federico Chiariotti
Reinforcement Learning for Durable Algorithmic Recourse
Algorithmic recourse gives individuals rejected by an automated decision system counterfactual explanations — recommendations meant to improve their odds on a future application. Prior work…
Contact: Marina Ceccon
Reinforcement Learning for Energy-Efficient Cloud Robotics
Cloud robotics offloads compute-heavy tasks — such as deep-network inference for semantic segmentation — from lightweight, compute-constrained robots to a cloud server, saving local compute…
With Massachusetts Institute of Technology (possibly remote)
Multi-Agent Reinforcement Learning for Compute Allocation in Processing Networks
Processing networks are multi-agent systems of smart sensors — each with sensing, compute, and communication capabilities — that refine data locally before sending it to a workstation for…
With Purdue University (possibly remote)
Graph Neural Network-Based Safe Multi-Robot Control
Multi-robot systems must satisfy strict safety requirements such as collision avoidance and minimum mutual distance. Control barrier functions (CBFs) are the standard tool for guaranteeing…
With National University of Singapore (possibly remote)
Resilient Distributed Federated Learning via Trust Observations
In distributed federated learning, agents collaboratively train machine-learning models by exchanging only model weights or gradients with their neighbours, without sharing raw data. This…
Contact: Gian Antonio Susto