Evaluation of user fatigue reduction through IEC rating-scale mapping

We evaluate the convergence speed of an Interactive Evolutionary Computation (IEC) using a rating-scale mapping for user fatigue reduction. First, we introduce the concept of mapping users' relative ratings to an "absolute scale "; this allows us to improve the performance of the IEC subjective evaluation characteristic predictor, which can in turn accelerate EC convergence and reduce user fatigue. Second, we experimentally evaluate the effectiveness of the proposed method using seven benchmark functions instead of a hunman user. The experimental results show that the convergence speed of an IEC using the proposed absolute rating data-trained predictor is much faster than an IEC using a conventional predictor trained using relative rating data.