Dipartimento di Scienze Statistiche

    Il Seminario di Dipartimento tenuto dal dott. Giorgio Grani si svolgerà in modalità telematica il giorno 25 NOVEMBRE p.v. con inizio alle ore 13.00 mediante accesso al seguente link: https://uniroma1.zoom.us/j/86128803817?pwd=QStzMzB0ZGJLWFc5Z3FWaUJ0L1Q1QT09  PASSCODE: 395069   Abstract: Scheduling problems have been one of the most challenging and studied problems in optimization in the last decades, and researchers from all over the world have tried a large and varied set of methods. This talk will tackle the job shop scheduling problem and the train dispatching problem via deep reinforcement learning. In particular, we will exploit the power of actor-critic algorithms and flexible neural structures, like LSTM and graph convolutional networks. For the job shop, we fight against CPLEX, whereas for the train dispatching we compare it to other learning-based heuristics.