kernDeepStackNet-package: Kernel deep stacking networks with random Fourier...

In kernDeepStackNet: Kernel Deep Stacking Networks

Description

Contains functions for estimation and model selection of kernel deep stacking networks (KDSNs). KDSNs are a supervised learning method, which can be used for continuous or binary responses. The model specification follows the approach of Huang et al. (2013), which is based on a series of kernel ridge regression models to random Fourier transformed input data.

The model selection includes model-based optimization of arbitrary loss functions. All help functions are also available for customized modeling, but it is recommended to use the higher level functions. The main functions are

• fitKDSN: Fits kernel deep stacking networks.

• tuneMboLevelCvKDSN: Selection of tuning parameters of kernel deep stacking networks with model-based optimization using cross-valdiation, arbitrary loss functions and pre-specified number of levels. All other tuning parameters are free to vary across the levels.

• tuneMboLevelGcvKDSN: Selection of tuning parameters of kernel deep stacking networks with model-based optimization using generalized cross validation score, arbitrary loss functions and pre-specified number of levels. All other tuning parameters are free to vary across the levels.

• tuneMboSharedCvKDSN: Selection of tuning parameters of kernel deep stacking networks with model-based optimization using cross validation or test set and arbitrary loss functions. The number of levels is included in tuning. All other tuning parameters are shared across the levels.

• tuneMboSharedSubsetKDSN: Selection of tuning parameters of kernel deep stacking networks with model-based optimization using test sets and arbitrary loss functions. The number of levels is included in tuning. All other tuning parameters are shared across the levels. In contrast to function tuneMboSharedCvKDSN an ensemble of (sparse) KDSNs is estimated.

For examples and further information, see the corresponding readme pages of the main functions.

Details

Package:	kernDeepStackNet
Type:	Package
Version:	2.0.2
Date:	2017-05-31
License:	GPL-3

Author(s)

Maintainer: Thomas Welchowski [e-mail]

Matthias Schmid [e-mail]

References

Thomas Welchowski, Matthias Schmid, (2016), A framework for parameter estimation and model selection in kernel deep stacking networks, Artificial Intelligence in Medicine, volume 70, pages 31-40

Po-Seng Huang and Li Deng and Mark Hasegawa-Johnson and Xiaodong He, (2013), Random Features for kernel deep convex network, Proceedings IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

Simon N. Wood, (2006), Generalized Additive Models: An Introduction with R, Taylor & Francis Group LLC

R. Brent, (1973), Algorithms for Minimization without Derivatives, Englewood Cliffs N.J.: Prentice-Hall

Donald R. Jones and Matthias Schonlau and William J. Welch, (1998), Efficient Global Optimization of Expensive Black-Box Functions, Journal of Global Optimization 13: pages 455-492

Krige DG, (1951), A Statistical Approach to Some Basic Mine Valuation Problems on the Witwatersrand, Journal of the Chemical, Metallurgical and Mining Society of South Africa, 52(6), 119-139

Olivier Roustant and David Ginsbourger and Yves Deville, (2012), DiceKriging, DiceOptim: Two R Packages for the Analysis of Computer Experiments by Kriging-Based Metamodeling and Optimization, Journal of Statistical Software, Volume 51, Issue 1

Michael Stein, (1987), Large Sample Properties of Simulations Using Latin Hypercube Sampling, Technometrics. 29, 143-151

Carl Edward Rasmussen and Christopher K. I. Williams, (2006), Gaussian Processes for Machine Learning, Massachusetts Institute of Technology

Jerome Friedman and Trevor Hastie and Rob Tibshirani, (2008), Regularization Paths for Generalized Linear Models via Coordinate Descent, Department of Statistics, Stanford University

Victor Picheny, David Ginsbourger, Yann Richet, (2012), Quantile-based optimization of Noisy Computer Experiments with Tunable Precision, HAL-archives-ouvertes.fr, hal-00578550v3

Nitish Srivastava, Geoffrey Hinton, Alex Krizhevsky et al., (2014), Dropout: A simple way to prevent neural networks from overfitting, Journal of Machine Learning Research, volume 15, pages 1929-1958

David Lopez-Paz and Philipp Hennig and Bernhard Sch\"olkopf, (2013), The randomized dependence coefficient, https://arxiv.org/, reference arXiv:1304.7717

Luca Scrucca, (2013), GA: A Package for Genetic Algorithms in R, Journal of Statistical Software, issue 4, volume 53, pages 1-37

Constantino Tsallis, Daniel A. Stariolo, (1996), Generalized simulated annealing, Elsevier, Physica A, volume 233, pages 395-406

kernDeepStackNet documentation built on May 2, 2019, 8:16 a.m.