Training Support Vector Machines (SVMs) to predict drugs concentrations is often difficult because of the high level of noise in the training data, due to various kinds of measurement errors. We apply RANdom SAmple Consensus (RANSAC) algorithm in this paper to solve this problem, enhancing the prediction accuracy by more than 40% in our particular case study. A personalized sample selection method is proposed to further improve the prediction result in most cases.
Friedrich Eisenbrand, Puck Elisabeth van Gerwen, Raimon Fabregat I De Aguilar-Amat
Michele Ceriotti, Federico Grasselli, Sanggyu Chong, Chiheb Ben Mahmoud
Alexandre Massoud Alahi, Saeed Saadatnejad, Taylor Ferdinand Mordan