Quantile regression neural networks: Implementation in R and application to precipitation downscaling

By Alex J. Cannon. Computational Geosciences, 2011

390 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Neural Networks in R Using the Stuttgart Neural Network Simulator: RSNNS (2012, methodology): “…version of the algorithm” presented by Anastasiadis et al. (2005), and a method for “the calculation of generalized weights”, which are not present in SNNS . monmlp The package monmlp (Cannon 2011a) implements a multi-layer perceptron with partial monotonicity constraints (Zhang and Zhang 1999).”
DeepSD: Generating High Resolution Climate Change Projections through Single Image Super-Resolution (2017, background): “Similar approaches, among others, include quantile regression neural networks [6], bayesian model averaging[45], and expanded”
Bias Correction of GCM Precipitation by Quantile Mapping: How Well Do Methods Preserve Changes in Quantiles and Extremes? (2015, neutral): “Similarly, more ﬂexible time-dependent quantile regression (Koenker and Schorfheide 1994; Cannon 2011) orparametricconditionaldensity estimators (Cannon2008, 2012), both of which can accommodate mixed discrete-continuous variables such as precipitation, may also perform better in this application.”
Probabilistic Forecasting with Spline Quantile Function RNNs (2019, methodology): “In previous work on quantile regression with neural networks, a smoothed version of the pinball loss— essentially an asymmetric version of the Huber loss— has been proposed, to avoid the non-di ↵ erentiability of the pinball loss function at zero (Cannon, 2011).”