From CAPTCHA to Commonsense: How Brain Can Teach Us About Artificial Intelligence

Despite the recent progress in AI powered by deep learning in solving narrow tasks, we are not close to human intelligence in its flexibility, versatility, and efficiency. Efficient learning and effective generalization come from inductive biases, and building Artificial General Intelligence (AGI) is an exercise in finding the right set of inductive biases that make fast learning possible while being general enough to be widely applicable in tasks that humans excel at. To make progress in AGI, we argue that we can look at the human brain for such inductive biases and principles of generalization. To that effect, we propose a strategy to gain insights from the brain by simultaneously looking at the world it acts upon and the computational framework to support efficient learning and generalization. We present a neuroscience-inspired generative model of vision as a case study for such approach and discuss some open problems about the path to AGI.

[1]  Bhaskara Marthi,et al.  A generative vision model that trains with high data efficiency and breaks text-based CAPTCHAs , 2017, Science.

[2]  Rolf A. Zwaan,et al.  Embodied Sentence Comprehension , 2005 .

[3]  Brian Cantwell Smith,et al.  The Promise of Artificial Intelligence: Reckoning and Judgment , 2019 .

[4]  L. Barsalou,et al.  Whither structured representation? , 1999, Behavioral and Brain Sciences.

[5]  Victor A. F. Lamme,et al.  Separate processing dynamics for texture elements, boundaries and surfaces in primary visual cortex of the macaque monkey. , 1999, Cerebral cortex.

[6]  Pieter R. Roelfsema,et al.  Object-based attention in the primary visual cortex of the macaque monkey , 1998, Nature.

[7]  Gary Marcus,et al.  Deep Learning: A Critical Appraisal , 2018, ArXiv.

[8]  S. Zeki,et al.  Modular Connections between Areas V2 and V4 of Macaque Monkey Visual Cortex , 1989, The European journal of neuroscience.

[9]  H. Chick,et al.  The photosensitizing action of buckwheat (Fagopyrum esculentum) , 1941, The Journal of physiology.

[10]  Yoshua Bengio,et al.  Scaling learning algorithms towards AI , 2007 .

[11]  W. Bair,et al.  Neural Coding for Shape and Texture in Macaque Area V4 , 2019, The Journal of Neuroscience.

[12]  L. Zhaoping.  Neural circuit models for computations in early visual cortex , 2011, Current Opinion in Neurobiology.

[13]  S Shipp.  Visual Processing: The odd couple , 1995, Current Biology.

[14]  Bernhard Schölkopf,et al.  Challenging Common Assumptions in the Unsupervised Learning of Disentangled Representations , 2018, ICML.

[15]  W. Freeman,et al.  Generalized Belief Propagation , 2000, NIPS.

[16]  Dileep George,et al.  How the brain might work: a hierarchical and temporal model for learning and recognition , 2008 .

[17]  Wolfgang Lehrach,et al.  Query Training: Learning and inference for directed and undirected graphical models , 2020, ArXiv.

[18]  S. Zeki,et al.  The Organization of Connections between Areas V5 and V1 in Macaque Monkey Visual Cortex , 1989, The European journal of neuroscience.

[19]  N. McLoughlin,et al.  Four Projection Streams from Primate V1 to the Cytochrome Oxidase Stripes of V2 , 2009, The Journal of Neuroscience.

[20]  D. C. Van Essen,et al.  Concurrent processing streams in monkey visual cortex , 1988, Trends in Neurosciences.

[21]  SchmidhuberJürgen.  Deep learning in neural networks , 2015 .

[22]  Joshua B. Tenenbaum,et al.  Approximate Bayesian Image Interpretation using Generative Probabilistic Graphics Programs , 2013, NIPS.

[23]  D. J. Felleman,et al.  Distributed hierarchical processing in the primate cerebral cortex. , 1991, Cerebral cortex.

[24]  Mélanie Frappier,et al.  The Book of Why: The New Science of Cause and Effect , 2018, Science.

[25]  Adam Santoro,et al.  Backpropagation and the brain , 2020, Nature Reviews Neuroscience.

[26]  J. Tsotsos.  What roles can attention play in recognition? , 2008, 2008 7th IEEE International Conference on Development and Learning.

[27]  Serge O. Dumoulin,et al.  Contour integration: Psychophysical, neurophysiological, and computational perspectives , 2014 .

[28]  Ernest Davis,et al.  Commonsense reasoning and commonsense knowledge in artificial intelligence , 2015, Commun. ACM.

[29]  Howard Hunt Pattee,et al.  Hierarchy Theory: The Challenge of Complex Systems , 1973 .

[30]  C. Gilbert,et al.  Top-down influences on visual processing , 2013, Nature Reviews Neuroscience.

[31]  G. Buzsáki.  The Brain from Inside Out , 2019 .

[32]  Guigang Zhang,et al.  Deep Learning , 2016, Int. J. Semantic Comput..

[33]  S. Hochstein,et al.  View from the Top Hierarchies and Reverse Hierarchies in the Visual System , 2002, Neuron.

[34]  Ohad Ben-Shahar,et al.  Geometrical Computations Explain Projection Patterns of Long-Range Horizontal Connections in Visual Cortex , 2004, Neural Computation.

[35]  K. Harris,et al.  Cortical connectivity and sensory coding , 2013, Nature.

[36]  Vernon B Mountcastle,et al.  Introduction. Computation in cortical columns. , 2003, Cerebral cortex.

[37]  Jürgen Schmidhuber,et al.  Deep learning in neural networks: An overview , 2014, Neural Networks.

[38]  Thomas Serre,et al.  Learning long-range spatial dependencies with horizontal gated-recurrent units , 2018, NeurIPS.

[39]  Paul G. Fahey,et al.  Cell type composition and circuit organization of clonally related excitatory neurons in the juvenile mouse neocortex , 2020, eLife.

[40]  David H. Wolpert,et al.  No free lunch theorems for optimization , 1997, IEEE Trans. Evol. Comput..

[41]  Henry Markram,et al.  A Novel Multiple Objective Optimization Framework for Constraining Conductance-Based Neuron Models by Experimental Data , 2007, Front. Neurosci..

[42]  Dileep George.  What can the brain teach us about building artificial intelligence? , 2017, The Behavioral and brain sciences.

[43]  J. M. Hupé,et al.  Cortical feedback improves discrimination between figure and background by V1, V2 and V3 neurons , 1998, Nature.

[44]  James J. DiCarlo,et al.  How Does the Brain Solve Visual Object Recognition? , 2012, Neuron.

[45]  D. Fitzpatrick,et al.  Orientation Selectivity and the Arrangement of Horizontal Connections in Tree Shrew Striate Cortex , 1997, The Journal of Neuroscience.

[46]  E. Callaway,et al.  Functional Streams and Local Connections of Layer 4C Neurons in Primary Visual Cortex of the Macaque Monkey , 1998, The Journal of Neuroscience.

[47]  N. McGlynn.  Thinking fast and slow. , 2014, Australian veterinary journal.

[48]  Uri Hasson,et al.  Direct Fit to Nature: An Evolutionary Perspective on Biological and Artificial Neural Networks , 2019, Neuron.

[49]  Stephen Grossberg,et al.  Filling-in the Forms: Surface and Boundary Interactions in Visual Cortex , 1998 .

[50]  W. Ashby,et al.  Every Good Regulator of a System Must Be a Model of That System , 1970 .

[51]  Illtyd Trethowan.  Causality , 1938 .

[52]  Anitha Pasupathy,et al.  Transformation of shape information in the ventral pathway , 2007, Current Opinion in Neurobiology.

[53]  T. S. Lee,et al.  Dynamics of subjective contour formation in the early visual cortex. , 2001, Proceedings of the National Academy of Sciences of the United States of America.

[54]  Tai Sing Lee,et al.  Hierarchical Bayesian inference in the visual cortex. , 2003, Journal of the Optical Society of America. A, Optics, image science, and vision.

[55]  Dileep George,et al.  Schema Networks: Zero-shot Transfer with a Generative Causal Model of Intuitive Physics , 2017, ICML.

[56]  P. Rakic.  Evolution of the neocortex: Perspective from developmental biology , 2010 .

[57]  Stuart Geman,et al.  Invariance and selectivity in the ventral visual pathway , 2006, Journal of Physiology-Paris.

[58]  J. Hochberg,et al.  Pictorial recognition as an unlearned ability: a study of one child's performance. , 1962, The American journal of psychology.

[59]  Karl J. Friston.  The free-energy principle: a unified brain theory? , 2010, Nature Reviews Neuroscience.

[60]  Steven W. Zucker,et al.  Third-Order Edge Statistics: Contour Continuation, Curvature, and Cortical Connections , 2013, NIPS.

[61]  T. S. Lee,et al.  A Bayesian framework for understanding texture segmentation in the primary visual cortex , 1995, Vision Research.

[62]  K. Ohki,et al.  Similarity of Visual Selectivity among Clonally Related Neurons in Visual Cortex , 2012, Neuron.

[63]  Tai Sing Lee,et al.  The Visual System's Internal Model of the World , 2015, Proceedings of the IEEE.

[64]  G. Schneider.  Brain Structure and Its Origins: in Development and in Evolution of Behavior and the Mind , 2014 .

[65]  Dileep George,et al.  Beyond imitation: Zero-shot task transfer on robots by learning concepts as cognitive programs , 2018, Science Robotics.

[66]  Eero P. Simoncelli,et al.  Natural image statistics and neural representation. , 2001, Annual review of neuroscience.

[67]  D. Hubel,et al.  Segregation of form, color, movement, and depth: anatomy, physiology, and perception. , 1988, Science.

[68]  Y. Dan,et al.  Clonally Related Visual Cortical Neurons Show Similar Stimulus Feature Selectivity , 2012, Nature.

[69]  Joshua B. Tenenbaum,et al.  Efficient inverse graphics in biological face processing , 2018, Science Advances.

[70]  Jean-Michel Morel,et al.  Cartoon+Texture Image Decomposition , 2011, Image Process. Line.

[71]  R. Weale.  Vision. A Computational Investigation Into the Human Representation and Processing of Visual Information. David Marr , 1983 .

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

[73]  Nikolaus Kriegeskorte,et al.  Recurrence is required to capture the representational dynamics of the human visual system , 2019, Proceedings of the National Academy of Sciences.

[74]  R. F Hess,et al.  Contour integration and cortical processing , 2003, Journal of Physiology-Paris.

[75]  Kate E. Jones,et al.  Temporal niche expansion in mammals from a nocturnal ancestor after dinosaur extinction , 2017, Nature Ecology & Evolution.

[76]  Ennio Mingolla,et al.  Neural dynamics of perceptual grouping: Textures, boundaries, and emergent segmentations , 1985 .

[77]  A. Grinvald,et al.  Relationship between intrinsic connections and functional architecture revealed by optical imaging and in vivo targeted biocytin injections in primate striate cortex. , 1993, Proceedings of the National Academy of Sciences of the United States of America.

[78]  S Grossberg,et al.  Neural dynamics of perceptual grouping: Textures, boundaries, and emergent segmentations , 1985, Perception & psychophysics.

[79]  A. Thomson,et al.  Functional Maps of Neocortical Local Circuitry , 2007, Front. Neurosci..

[80]  Judea Pearl,et al.  Probabilistic reasoning in intelligent systems - networks of plausible inference , 1991, Morgan Kaufmann series in representation and reasoning.

[81]  Venkat Chandrasekaran,et al.  Surflets: a sparse representation for multidimensional functions containing smooth discontinuities , 2004, International Symposium onInformation Theory, 2004. ISIT 2004. Proceedings..