Selective Perception as a Mechanism to Adapt Agents to the Environment: An Evolutionary Approach

Rapid advancement of machine learning makes it possible to consider large amounts of data to learn from. Learning agents may get data ranging on real intervals directly from the environment they interact with, in a process usually time expensive. To improve learning and manage these data, approximated models and memory mechanisms are adopted. In most of the implementations of reinforcement learning facing this type of data, approximation is obtained by neural networks and the process of drawing information from data is mediated by a shortterm memory that stores the previous experiences for additional relearning, to speed-up the learning process, mimicking what is done by people. In this paper, we are proposing a novel computational approach able to selectively filter the information, or cognitive load, for the agent's short-term memory, thus emulating the attention mechanism characteristic of human perception. In this work, we use genetic algorithms in order to evolve the most efficient attention filter mechanism that would be able to provide the agent with an optimal perception for a specific environment by discriminating which experiences are valuable for the learning process. This approach can evolve a filter which can able to provide an optimal cognitive load of the experiences entering in the agent's short-term memory of a limited capacity. The evolved sampling dynamics can also lead to the emergence of intrinsically motivated curiosity.