Enviar Registro por Correo electrónico: Adaptive representations for reinforcement learning /