A Memory-Augmented Neural Network Model of Abstract Rule Learning

Human intelligence is characterized by a remarkable ability to infer abstract rules from experience and apply these rules to novel domains. As such, designing neural network algorithms with this capacity is an important step toward the development of deep learning systems with more human-like intelligence. However, doing so is a major outstanding challenge, one that some argue will require neural networks to use explicit symbol-processing mechanisms. In this work, we focus on neural networks' capacity for arbitrary role-filler binding, the ability to associate abstract "roles" to context-specific "fillers," which many have argued is an important mechanism underlying the ability to learn and apply rules abstractly. Using a simplified version of Raven's Progressive Matrices, a hallmark test of human intelligence, we introduce a sequential formulation of a visual problem-solving task that requires this form of binding. Further, we introduce the Emergent Symbol Binding Network (ESBN), a recurrent neural network model that learns to use an external memory as a binding mechanism. This mechanism enables symbol-like variable representations to emerge through the ESBN's training process without the need for explicit symbol-processing machinery. We empirically demonstrate that the ESBN successfully learns the underlying abstract rule structure of our task and perfectly generalizes this rule structure to novel fillers.

[1]  Dedre Gentner,et al.  Structure-Mapping: A Theoretical Framework for Analogy , 1983, Cogn. Sci..

[2]  Pietro Lio,et al.  Generalisable Relational Reasoning With Comparators in Low-Dimensional Manifolds , 2020, ArXiv.

[3]  Daan Wierstra,et al.  Meta-Learning with Memory-Augmented Neural Networks , 2016, ICML.

[4]  Kenneth D. Forbus,et al.  Modeling Visual Problem Solving as Analogical Reasoning , 2017, Psychological review.

[5]  K. Holyoak Analogy and Relational Reasoning , 2012 .

[6]  Jian Sun,et al.  Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification , 2015, 2015 IEEE International Conference on Computer Vision (ICCV).

[7]  Felix Hill,et al.  Learning to Make Analogies by Contrasting Abstract Relational Structure , 2019, ICLR.

[8]  G. J. Robertson Raven's Progressive Matrices , 2010 .

[9]  Razvan Pascanu,et al.  A simple neural network module for relational reasoning , 2017, NIPS.

[10]  Demis Hassabis,et al.  Neural Episodic Control , 2017, ICML.

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

[12]  K. Holyoak,et al.  A symbolic-connectionist theory of relational inference and generalization. , 2003, Psychological review.

[13]  Caswell Barry,et al.  The Tolman-Eichenbaum Machine: Unifying Space and Relational Memory through Generalization in the Hippocampal Formation , 2019, Cell.

[14]  Brian Falkenhainer,et al.  The Structure-Mapping Engine: Algorithm and Examples , 1989, Artif. Intell..

[15]  Thomas Martinetz,et al.  Multi-layer Relation Networks , 2018, ArXiv.

[16]  Alex Graves,et al.  Neural Turing Machines , 2014, ArXiv.

[17]  Jonathan D. Cohen,et al.  Indirection and symbol-like processing in the prefrontal cortex and basal ganglia , 2013, Proceedings of the National Academy of Sciences.

[18]  Felix Hill,et al.  Measuring abstract reasoning in neural networks , 2018, ICML.

[19]  Daniel M. Cer,et al.  Neural mechanisms of binding in the hippocampus and neocortex: insights from computational models , 2006, Handbook of Binding and Memory.

[20]  E John,et al.  A Process Model of Human Transitive Inference , 2001 .

[21]  Geoffrey E. Hinton,et al.  Using Fast Weights to Attend to the Recent Past , 2016, NIPS.

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

[23]  James L. McClelland,et al.  Why there are complementary learning systems in the hippocampus and neocortex: insights from the successes and failures of connectionist models of learning and memory. , 1995, Psychological review.

[24]  Yixin Zhu,et al.  Learning Perceptual Inference by Contrasting , 2019, NeurIPS.

[25]  Yoshua Bengio,et al.  Understanding the difficulty of training deep feedforward neural networks , 2010, AISTATS.

[26]  Xianglong Liu,et al.  Hierarchical Rule Induction Network for Abstract Visual Reasoning , 2020, ArXiv.

[27]  Pietro Liò,et al.  Abstract Diagrammatic Reasoning with Multiplex Graph Networks , 2020, ICLR.

[28]  Jimmy Ba,et al.  The Scattering Compositional Learner: Discovering Objects, Attributes, Relationships in Analogical Reasoning , 2020, ArXiv.

[29]  Joshua B. Tenenbaum,et al.  Building machines that learn and think like people , 2016, Behavioral and Brain Sciences.

[30]  Thomas Martinetz,et al.  Solving Raven’s Progressive Matrices with Multi-Layer Relation Networks , 2020, 2020 International Joint Conference on Neural Networks (IJCNN).

[31]  J. Raven,et al.  Raven Progressive Matrices , 2003 .

[32]  Lukasz Kaiser,et al.  Attention is All you Need , 2017, NIPS.

[33]  Marco Baroni,et al.  Still not systematic after all these years: On the compositional skills of sequence-to-sequence recurrent networks , 2017, ICLR 2018.

[34]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[35]  C. E. Bethell-Fox,et al.  Adaptive reasoning: Componential and eye movement analysis of geometric analogy performance ☆ , 1984 .

[36]  John E. Hummel,et al.  Distributed representations of structure: A theory of analogical access and mapping. , 1997 .

[37]  M A Just,et al.  From the SelectedWorks of Marcel Adam Just 1990 What one intelligence test measures : A theoretical account of the processing in the Raven Progressive Matrices Test , 2016 .

[38]  Feng Gao,et al.  RAVEN: A Dataset for Relational and Analogical Visual REasoNing , 2019, 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).