State-Reification Networks: Improving Generalization by Modeling the Distribution of Hidden Representations

Machine learning promises methods that generalize well from finite labeled data. However, the brittleness of existing neural net approaches is revealed by notable failures, such as the existence of adversarial examples that are misclassified despite being nearly identical to a training example, or the inability of recurrent sequence-processing nets to stay on track without teacher forcing. We introduce a method, which we refer to as \emph{state reification}, that involves modeling the distribution of hidden states over the training data and then projecting hidden states observed during testing toward this distribution. Our intuition is that if the network can remain in a familiar manifold of hidden space, subsequent layers of the net should be well trained to respond appropriately. We show that this state-reification method helps neural nets to generalize better, especially when labeled data are sparse, and also helps overcome the challenge of achieving robust generalization with adversarial training.

[1]  A. Reber.  Implicit learning of artificial grammars , 1967 .

[2]  S. Boll,et al.  Suppression of acoustic noise in speech using spectral subtraction , 1979 .

[3]  J J Hopfield,et al.  Neurons with graded response have collective computational properties like those of two-state neurons. , 1984, Proceedings of the National Academy of Sciences of the United States of America.

[4]  Geoffrey E. Hinton.  Mapping Part-Whole Hierarchies into Connectionist Networks , 1990, Artif. Intell..

[5]  Yann LeCun,et al.  Tangent Prop - A Formalism for Specifying Selected Invariances in an Adaptive Network , 1991, NIPS.

[6]  Jude W. Shavlik,et al.  Learning Symbolic Rules Using Artificial Neural Networks , 1993, ICML.

[7]  Pascal Koiran.  Dynamics of Discrete Time, Continuous State Hopfield Networks , 1994, Neural Computation.

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

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

[10]  Sepp Hochreiter,et al.  The Vanishing Gradient Problem During Learning Recurrent Neural Nets and Problem Solutions , 1998, Int. J. Uncertain. Fuzziness Knowl. Based Syst..

[11]  John J. Hopfield,et al.  Neural networks and physical systems with emergent collective computational abilities , 1999 .

[12]  M. Mozer.  Attractor Networks , 2000 .

[13]  M. Masson.  Using confidence intervals for graphically based data interpretation. , 2003, Canadian journal of experimental psychology = Revue canadienne de psychologie experimentale.

[14]  Hava T. Siegelmann,et al.  Analog-symbolic memory that tracks via reconsolidation , 2008 .

[15]  Pascal Vincent,et al.  Representation Learning: A Review and New Perspectives , 2012, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[16]  Yoshua Bengio,et al.  Regularized Auto-Encoders Estimate Local Statistics , 2012, ICLR.

[17]  Yoshua Bengio,et al.  Better Mixing via Deep Representations , 2012, ICML.

[18]  Joan Bruna,et al.  Intriguing properties of neural networks , 2013, ICLR.

[19]  Luca Rigazio,et al.  Towards Deep Neural Network Architectures Robust to Adversarial Examples , 2014, ICLR.

[20]  Jonathon Shlens,et al.  Explaining and Harnessing Adversarial Examples , 2014, ICLR.

[21]  Samy Bengio,et al.  Scheduled Sampling for Sequence Prediction with Recurrent Neural Networks , 2015, NIPS.

[22]  Nikos Komodakis,et al.  Wide Residual Networks , 2016, BMVC.

[23]  Ananthram Swami,et al.  Distillation as a Defense to Adversarial Perturbations Against Deep Neural Networks , 2015, 2016 IEEE Symposium on Security and Privacy (SP).

[24]  Renjie Liao,et al.  Learning Deep Parsimonious Representations , 2016, NIPS.

[25]  Jian Sun,et al.  Identity Mappings in Deep Residual Networks , 2016, ECCV.

[26]  Yang Song,et al.  Improving the Robustness of Deep Neural Networks via Stability Training , 2016, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[27]  Alexandros G. Dimakis,et al.  The Robust Manifold Defense: Adversarial Training using Generative Models , 2017, ArXiv.

[28]  Yoshua Bengio.  The Consciousness Prior , 2017, ArXiv.

[29]  Hao Chen,et al.  MagNet: A Two-Pronged Defense against Adversarial Examples , 2017, CCS.

[30]  Yoshua Bengio,et al.  Improving Generative Adversarial Networks with Denoising Feature Matching , 2016, ICLR.

[31]  Martín Abadi,et al.  Adversarial Patch , 2017, ArXiv.

[32]  Jun Zhu,et al.  Towards Robust Detection of Adversarial Examples , 2017, NeurIPS.

[33]  Ioannis Mitliagkas,et al.  Fortified Networks: Improving the Robustness of Deep Networks by Modeling the Manifold of Hidden Representations , 2018, ArXiv.

[34]  Roberto Caldelli,et al.  Adversarial Examples Detection in Features Distance Spaces , 2018, ECCV Workshops.

[35]  Kibok Lee,et al.  Training Confidence-calibrated Classifiers for Detecting Out-of-Distribution Samples , 2017, ICLR.

[36]  Michael C. Mozer,et al.  State-Denoised Recurrent Neural Networks , 2018, ArXiv.

[37]  David A. Wagner,et al.  Obfuscated Gradients Give a False Sense of Security: Circumventing Defenses to Adversarial Examples , 2018, ICML.

[38]  Aleksander Madry,et al.  Towards Deep Learning Models Resistant to Adversarial Attacks , 2017, ICLR.

[39]  Aleksander Madry,et al.  Adversarially Robust Generalization Requires More Data , 2018, NeurIPS.

[40]  Colin Raffel,et al.  Thermometer Encoding: One Hot Way To Resist Adversarial Examples , 2018, ICLR.

[41]  Yanjun Qi,et al.  Feature Squeezing: Detecting Adversarial Examples in Deep Neural Networks , 2017, NDSS.

[42]  Yi Zhang,et al.  Stronger generalization bounds for deep nets via a compression approach , 2018, ICML.

[43]  Xiaolin Hu,et al.  Defense Against Adversarial Attacks Using High-Level Representation Guided Denoiser , 2017, 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.

[44]  Kibok Lee,et al.  A Simple Unified Framework for Detecting Out-of-Distribution Samples and Adversarial Attacks , 2018, NeurIPS.

[45]  Martin Wattenberg,et al.  Adversarial Spheres , 2018, ICLR.

[46]  Dan Klein,et al.  Learning with Latent Language , 2017, NAACL.

[47]  Rama Chellappa,et al.  Defense-GAN: Protecting Classifiers Against Adversarial Attacks Using Generative Models , 2018, ICLR.