A Learning-Based Approach to Artificial Sensory Feedback