On Fast Dropout and its Applicability to Recurrent Networks

Recurrent Neural Networks (RNNs) are rich models for the processing of sequential data. Recent work on advancing the state of the art has been focused on the optimization or modelling of RNNs, mostly motivated by adressing the problems of the vanishing and exploding gradients. The control of overfitting has seen considerably less attention. This paper contributes to that by analyzing fast dropout, a recent regularization method for generalized linear models and neural networks from a back-propagation inspired perspective. We show that fast dropout implements a quadratic form of an adaptive, per-parameter regularizer, which rewards large weights in the light of underfitting, penalizes them for overconfident predictions and vanishes at minima of an unregularized training loss. The derivatives of that regularizer are exclusively based on the training error signal. One consequence of this is the absense of a global weight attractor, which is particularly appealing for RNNs, since the dynamics are not biased towards a certain regime. We positively test the hypothesis that this improves the performance of RNNs on four musical data sets.

[1]  Charles R. Johnson,et al.  Matrix analysis , 1985, Statistical Inference for Engineers and Data Scientists.

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

[3]  Sepp Hochreiter,et al.  Untersuchungen zu dynamischen neuronalen Netzen , 1991 .

[4]  Heekuck Oh,et al.  Neural Networks for Pattern Recognition , 1993, Adv. Comput..

[5]  Yoshua Bengio,et al.  Learning long-term dependencies with gradient descent is difficult , 1994, IEEE Trans. Neural Networks.

[6]  Jürgen Schmidhuber,et al.  Long Short-Term Memory , 1997, Neural Computation.

[7]  Jeffrey K. Uhlmann,et al.  New extension of the Kalman filter to nonlinear systems , 1997, Defense, Security, and Sensing.

[8]  Barbara Hammer,et al.  On the approximation capability of recurrent neural networks , 2000, Neurocomputing.

[9]  E. Lehmann.  Elements of large-sample theory , 1998 .

[10]  Herbert Jaeger,et al.  Adaptive Nonlinear System Identification with Echo State Networks , 2002, NIPS.

[11]  Nasser M. Nasrabadi,et al.  Pattern Recognition and Machine Learning , 2006, Technometrics.

[12]  Jürgen Schmidhuber,et al.  Unconstrained On-line Handwriting Recognition with Recurrent Neural Networks , 2007, NIPS.

[13]  A. Graves,et al.  Unconstrained Online Handwriting Recognition with Recurrent Neural Networks , 2007 .

[14]  Geoffrey E. Hinton,et al.  The Recurrent Temporal Restricted Boltzmann Machine , 2008, NIPS.

[15]  Mert Bay,et al.  Evaluation of Multiple-F0 Estimation and Tracking Systems , 2009, ISMIR.

[16]  Yoram Singer,et al.  Adaptive Subgradient Methods for Online Learning and Stochastic Optimization , 2011, J. Mach. Learn. Res..

[17]  Lukás Burget,et al.  Recurrent neural network based language model , 2010, INTERSPEECH.

[18]  Ilya Sutskever,et al.  Learning Recurrent Neural Networks with Hessian-Free Optimization , 2011, ICML.

[19]  Alex Graves,et al.  Practical Variational Inference for Neural Networks , 2011, NIPS.

[20]  Pascal Vincent,et al.  The Manifold Tangent Classifier , 2011, NIPS.

[21]  Geoffrey E. Hinton,et al.  Generating Text with Recurrent Neural Networks , 2011, ICML.

[22]  Nitish Srivastava,et al.  Improving neural networks by preventing co-adaptation of feature detectors , 2012, ArXiv.

[23]  Yoshua Bengio,et al.  Random Search for Hyper-Parameter Optimization , 2012, J. Mach. Learn. Res..

[24]  Geoffrey E. Hinton,et al.  Speech recognition with deep recurrent neural networks , 2013, 2013 IEEE International Conference on Acoustics, Speech and Signal Processing.

[25]  Maneesh Sahani,et al.  Regularization and nonlinearities for neural language models: when are they needed? , 2013, ArXiv.

[26]  Geoffrey E. Hinton,et al.  On rectified linear units for speech processing , 2013, 2013 IEEE International Conference on Acoustics, Speech and Signal Processing.

[27]  Razvan Pascanu,et al.  On the difficulty of training recurrent neural networks , 2012, ICML.

[28]  Alex Graves,et al.  Generating Sequences With Recurrent Neural Networks , 2013, ArXiv.

[29]  Christian Osendorfer,et al.  Training Neural Networks with Implicit Variance , 2013, ICONIP.

[30]  Yoshua Bengio,et al.  High-dimensional sequence transduction , 2012, 2013 IEEE International Conference on Acoustics, Speech and Signal Processing.

[31]  Geoffrey E. Hinton,et al.  On the importance of initialization and momentum in deep learning , 2013, ICML.

[32]  Sida I. Wang,et al.  Dropout Training as Adaptive Regularization , 2013, NIPS.

[33]  Razvan Pascanu,et al.  Advances in optimizing recurrent networks , 2012, 2013 IEEE International Conference on Acoustics, Speech and Signal Processing.

[34]  Christopher D. Manning,et al.  Fast dropout training , 2013, ICML.

[35]  Geoffrey E. Hinton,et al.  Training Recurrent Neural Networks , 2013 .

[36]  Razvan Pascanu,et al.  How to Construct Deep Recurrent Neural Networks , 2013, ICLR.