access_time 27 de janeiro de 2020 às 10:00 até 27 de janeiro de 2020 às 11:00
place Room 2-N7.1, IST, Taguspark
This paper addresses the problem of ad hoc teamwork (where an agent is paired with a team of agents, all sharing a common goal) without a reinforcement signal and visible teammates' actions. We extend the work of Melo and Sardinha [12] to sequential problems with no action observability and conduct an empirical evaluation (N=1000, confidence level of 99\%) comparing our approach against two other solutions. We show that our approach is reliable in a virtual environment where two agents must assist each other in completing a given task.
local_offer Tópicos de Investigação
person Candidato: João Manuel Godinho Ribeiro
supervisor_account Orientador: Prof. Francisco Melo / Prof. José Alberto Sardinha