Neural network embedding of the over-dispersed Poisson reserving model

ABSTRACT The main idea of this paper is to embed a classical actuarial regression model into a neural network architecture. This nesting allows us to learn model structure beyond the classical actuarial regression model if we use as starting point of the neural network calibration exactly the classical actuarial model. Such models can be fitted efficiently which allows us to explore bootstrap methods for prediction uncertainty. As an explicit example, we consider the cross-classified over-dispersed Poisson model for general insurance claims reserving. We demonstrate how this model can be improved by neural network features.

[1]  George Cybenko,et al.  Approximation by superpositions of a sigmoidal function , 1989, Math. Control. Signals Syst..

[2]  Andrea Gabrielli,et al.  A Neural Network Boosted Double Over-Dispersed Poisson Claims Reserving Model , 2019, SSRN Electronic Journal.

[3]  Sergey Ioffe,et al.  Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift , 2015, ICML.

[4]  Geoffrey E. Hinton,et al.  Learning representations by back-propagating errors , 1986, Nature.

[5]  Geoffrey E. Hinton,et al.  Deep Learning , 2015, Nature.

[6]  E. L. Lehmann,et al.  Theory of point estimation , 1950 .

[7]  Yoshua Bengio,et al.  A Neural Probabilistic Language Model , 2003, J. Mach. Learn. Res..

[8]  Mario V. Wuthrich,et al.  Stochastic Claims Reserving Manual: Advances in Dynamic Modeling , 2015 .

[9]  Jürgen Schmidhuber,et al.  Deep learning in neural networks: An overview , 2014, Neural Networks.

[10]  P. England,et al.  Analytic and bootstrap estimates of prediction errors in claims reserving , 1999 .

[11]  Eric R. Ziegel,et al.  Generalized Linear Models , 2002, Technometrics.

[12]  M. Wüthrich,et al.  An Individual Claims History Simulation Machine , 2018 .

[13]  Alois Gisler,et al.  Credibility for the Chain Ladder Reserving Method , 2008 .

[14]  Guangyuan Gao,et al.  Bayesian Chain Ladder Models , 2018 .

[15]  T. Mack.  Distribution-free Calculation of the Standard Error of Chain Ladder Reserve Estimates , 1993, ASTIN Bulletin.

[16]  Andrea Gabrielli,et al.  An Individual Claims History Simulation Machine , 2018 .

[17]  L. Rüschendorf,et al.  COMPLETENESS IN LOCATION FAMILIES , 2008 .

[18]  Kurt Hornik,et al.  Multilayer feedforward networks are universal approximators , 1989, Neural Networks.

[19]  Ronald Richman,et al.  AI in Actuarial Science , 2018 .

[20]  P. England,et al.  Stochastic Claims Reserving in General Insurance , 2002, British Actuarial Journal.

[21]  Erhard Kremer.  Einführung in die Versicherungsmathematik , 1985 .

[22]  Mario V. Wuthrich,et al.  A neural network extension of the Lee–Carter model to multiple populations , 2018, Annals of Actuarial Science.

[23]  Guigang Zhang,et al.  Deep Learning , 2016, Int. J. Semantic Comput..

[24]  Nitish Srivastava,et al.  Dropout: a simple way to prevent neural networks from overfitting , 2014, J. Mach. Learn. Res..

[25]  Thomas Mack,et al.  A Simple Parametric Model for Rating Automobile Insurance or Estimating IBNR Claims Reserves , 1991, ASTIN Bulletin.

[26]  Michael Merz,et al.  EDITORIAL: YES, WE CANN! , 2018, ASTIN Bulletin.

[27]  Richard Verrall,et al.  A Stochastic Model Underlying the Chain-Ladder Technique , 1998, British Actuarial Journal.