10.1371/journal.pone.0138507.g002 Paulo S. G. de Mattos Neto Paulo S. G. de Mattos Neto George D. C. Cavalcanti George D. C. Cavalcanti Francisco Madeiro Francisco Madeiro Tiago A. E. Ferreira Tiago A. E. Ferreira Training architecture of the proposed system. Public Library of Science 2015 ga Artificial neural networks approach correction PM 10 concentration series High PM concentration hs residuals modeling fitness function ann PM forecasters forecasting method performance Experimental results show PM time series 2015-09-28 03:21:04 Figure https://plos.figshare.com/articles/figure/_Training_architecture_of_the_proposed_system_/1558898 <p>Training architecture of the proposed system.</p>