Convolutional Bipartite Attractor Networks

In human perception and cognition, a fundamental operation that brains perform is interpretation: constructing coherent neural states from noisy, incomplete, and intrinsically ambiguous evidence. The problem of interpretation is well matched to an early and often overlooked architecture, the attractor network---a recurrent neural net that performs constraint satisfaction, imputation of missing features, and clean up of noisy data via energy minimization dynamics. We revisit attractor nets in light of modern deep learning methods and propose a convolutional bipartite architecture with a novel training loss, activation function, and connectivity constraints. We tackle larger problems than have been previously explored with attractor nets and demonstrate their potential for image completion and super-resolution. We argue that this architecture is better motivated than ever-deeper feedforward models and is a viable alternative to more costly sampling-based generative methods on a range of supervised and unsupervised tasks.

[1]  Narendra Ahuja,et al.  Fast and Accurate Image Super-Resolution with Deep Laplacian Pyramid Networks , 2017, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[2]  Lisa Zhang,et al.  Reviving and Improving Recurrent Back-Propagation , 2018, ICML.

[3]  M. Kutas,et al.  Interactions between sentence context and word frequencyinevent-related brainpotentials , 1990, Memory & cognition.

[4]  Fu Jie Huang,et al.  A Tutorial on Energy-Based Learning , 2006 .

[5]  J J Hopfield,et al.  Neural networks and physical systems with emergent collective computational abilities. , 1982, Proceedings of the National Academy of Sciences of the United States of America.

[6]  Tian Han,et al.  Divergence Triangle for Joint Training of Generator Model, Energy-Based Model, and Inferential Model , 2018, 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[7]  James L. McClelland,et al.  Parallel distributed processing: explorations in the microstructure of cognition, vol. 1: foundations , 1986 .

[8]  Tafsir Thiam,et al.  The Boltzmann machine , 1999, IJCNN'99. International Joint Conference on Neural Networks. Proceedings (Cat. No.99CH36339).

[9]  Jitendra Malik,et al.  A database of human segmented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics , 2001, Proceedings Eighth IEEE International Conference on Computer Vision. ICCV 2001.

[10]  Ken Perlin,et al.  An image synthesizer , 1988 .

[11]  Richard S. Zemel,et al.  Localist Attractor Networks , 2001, Neural Computation.

[12]  Alex Graves,et al.  The Kanerva Machine: A Generative Distributed Memory , 2018, ICLR.

[13]  Luping Shi,et al.  Memory Dynamics in Attractor Networks , 2015, Comput. Intell. Neurosci..

[14]  Kyoung Mu Lee,et al.  Deeply-Recursive Convolutional Network for Image Super-Resolution , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[15]  Surya Ganguli,et al.  Task-Driven Convolutional Recurrent Models of the Visual System , 2018, NeurIPS.

[16]  Daniel J. Amit,et al.  Modeling brain function: the world of attractor neural networks, 1st Edition , 1989 .

[17]  Yoshua Bengio,et al.  Gradient Flow in Recurrent Nets: the Difficulty of Learning Long-Term Dependencies , 2001 .

[18]  Jian Yang,et al.  Image Super-Resolution via Deep Recursive Residual Network , 2017, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[19]  Philipp Sterzer,et al.  A neural basis for inference in perceptual ambiguity , 2007, Proceedings of the National Academy of Sciences.

[20]  Geoffrey E. Hinton,et al.  Exponential Family Harmoniums with an Application to Information Retrieval , 2004, NIPS.

[21]  Geoffrey E. Hinton,et al.  Reducing the Dimensionality of Data with Neural Networks , 2006, Science.

[22]  Joshua B. Tenenbaum,et al.  Human-level concept learning through probabilistic program induction , 2015, Science.

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

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

[25]  Honglak Lee,et al.  Convolutional deep belief networks for scalable unsupervised learning of hierarchical representations , 2009, ICML '09.

[26]  Rishidev Chaudhuri,et al.  Associative content-addressable networks with exponentially many robust stable states , 2017, ArXiv.

[27]  Pineda,et al.  Generalization of back-propagation to recurrent neural networks. , 1987, Physical review letters.

[28]  John J. Hopfield,et al.  Dense Associative Memory for Pattern Recognition , 2016, NIPS.

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

[30]  Eero P. Simoncelli,et al.  Image quality assessment: from error visibility to structural similarity , 2004, IEEE Transactions on Image Processing.

[31]  Yang Lu,et al.  A Theory of Generative ConvNet , 2016, ICML.

[32]  Igor Mordatch,et al.  Implicit Generation and Generalization with Energy Based Models , 2018 .

[33]  Yoshua Bengio,et al.  Extracting and composing robust features with denoising autoencoders , 2008, ICML '08.

[34]  Max Welling,et al.  Auto-Encoding Variational Bayes , 2013, ICLR.

[35]  Samy Bengio,et al.  Density estimation using Real NVP , 2016, ICLR.

[36]  Richard S. Sutton,et al.  Learning to predict by the methods of temporal differences , 1988, Machine Learning.

[37]  Thomas S. Huang,et al.  Image Super-Resolution Via Sparse Representation , 2010, IEEE Transactions on Image Processing.

[38]  Emile H. L. Aarts,et al.  Boltzmann machines , 1998 .

[39]  Haizhou Li,et al.  Continuous attractors of discrete-time recurrent neural networks , 2012, Neural Computing and Applications.

[40]  M. Mozer Attractor Networks , 2000 .

[41]  Edith Kaan,et al.  The sentence wrap-up dogma , 2018, Cognition.

[42]  James L. McClelland,et al.  An interactive activation model of context effects in letter perception: I. An account of basic findings. , 1981 .

[43]  Narendra Ahuja,et al.  Single image super-resolution from transformed self-exemplars , 2015, 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[44]  Aline Roumy,et al.  Low-Complexity Single-Image Super-Resolution based on Nonnegative Neighbor Embedding , 2012, BMVC.

[45]  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.

[46]  James J. DiCarlo,et al.  Evidence that recurrent circuits are critical to the ventral stream’s execution of core object recognition behavior , 2018, Nature Neuroscience.

[47]  Yan Wu,et al.  Learning Attractor Dynamics for Generative Memory , 2018, NeurIPS.

[48]  Ioannis Mitliagkas,et al.  State-Reification Networks: Improving Generalization by Modeling the Distribution of Hidden Representations , 2019, ICML.

[49]  Michael Elad,et al.  On Single Image Scale-Up Using Sparse-Representations , 2010, Curves and Surfaces.

[50]  Luís B. Almeida,et al.  A learning rule for asynchronous perceptrons with feedback in a combinatorial environment , 1990 .

[51]  David Cox,et al.  Recurrent computations for visual pattern completion , 2017, Proceedings of the National Academy of Sciences.