Configuration of the Deep Neural Network Hyperparameters for the Hypsometric Modeling of the Guazuma crinita Mart. in the Peruvian Amazon

The Guazuma crinita Mart. is a dominant species of great economic importance for the inhabitants Peruvian Amazon, standing out its rapid growth and being harvested at an early age. Understanding vertical challenge that researchers have continued to study using different hypsometric modeling techniques. Currently, machine learning techniques, especially artificial neural networks, revolutionized forest management, obtaining more accurate predictions; it because we understand utmost adapt, evaluate apply these methods in this large areas. objective was build efficiency use deep network prediction total height from large-scale continuous inventory. To do this, explore configurations hidden layer hyperparameters define variables according function HT = f(x) where as output variable x input variable(s). Under criterion, established three relationships: based on diameter breast (DBH), (i) f(DBH); DBH Age, (ii) f(DBH, Age) DBH, Age Agroclimatic variables, (iii) Agroclimatology), respectively. In total, 24 configuration models were each function, concluding technique presents satisfactory performance predictions areas, agroclimatic with validation RMSE 0.70, MAE 0.50, bias% −0.09 VAR 0.49, showed better accuracy than others.