Alessandro
Bordonali

M.Sc. Computer Science

Università degli Studi di Milano

Thesis: TBD

B.Sc. Computer Science

Università degli Studi di Milano

Thesis: Definizione e valutazione sperimentale di algoritmi per multi-agent path planning in applicazioni di monitoraggio ambientale con sistemi multi-robot

B.Sc. Thesis

Università degli Studi di Milano

Defined and experimentally evaluated path planning algorithms for multi-agent environmental monitoring with multi-robot systems.

F1 Racing with Policy Gradient

Trained an autonomous agent to race an F1 car on a custom continuous MDP (position, velocity → acceleration, steering) using policy gradient methods. Compared MLPs of varying depth and width on convergence speed and lap time. UniMI Reinforcement Learning course (Cesa-Bianchi, Papini, Ferrara).

Hedge vs. AdaHedge

Empirical study of fixed vs. adaptive learning rates in the prediction with expert advice setting. Compared Hedge and AdaHedge (van Erven et al., 2011) across stochastic, adversarial, and low-gap regimes — AdaHedge matched optimal regret without manual rate tuning.

pp-seg

Semantic segmentation project — notebooks and model utilities for training and evaluating segmentation pipelines. Python, Jupyter.