Learning in Cognitive Architectures
The line of learning in cognitive architectures aims to advance the state of the art of artificial intelligence through the use of learning in cognitive agents. In particular, the challenges involved in using training models on mobile devices in order to improve the user experience with these devices will be analyzed. In this scenario, different learning paradigms can be used. However, given the typical characteristics of the problem where the cognitive agent must reflect the experiences of each user, the training of models based on Reinforcement Learning (AR) should be more evident. Among the challenges in the area are: the design of reinforcement functions that are capable of implicitly or explicitly evaluating the quality of the agent's decision-making and the way in which these agents are trained, since models based on AR require numerous iterations to converge. Therefore, the transfer of learning, offline RL and the construction of simulated scenarios can be of great importance.