Unsupervised Grounding of Plannable First-Order Logic Representation from Images

Recently, there is an increasing interest in obtaining the relational structures of the environment in the Reinforcement Learning community. However, the resulting "relations" are not the discrete, logical predicates compatible to the symbolic reasoning such as classical planning or goal recognition. Meanwhile, Latplan (Asai and Fukunaga 2018) bridged the gap between deep-learning perceptual systems and symbolic classical planners. One key component of the system is a Neural Network called State AutoEncoder (SAE), which encodes an image-based input into a propositional representation compatible to classical planning. To get the best of both worlds, we propose First-Order State AutoEncoder, an unsupervised architecture for grounding the first-order logic predicates and facts. Each predicate models a relationship between objects by taking the interpretable arguments and returning a propositional value. In the experiment using 8-Puzzle and a photo-realistic Blocksworld environment, we show that (1) the resulting predicates capture the interpretable relations (e.g. spatial), (2) they help obtaining the compact, abstract model of the environment, and finally, (3) the resulting model is compatible to symbolic classical planning.

[1]  E. Gumbel.  Statistical Theory of Extreme Values and Some Practical Applications : A Series of Lectures , 1954 .

[2]  Peter Norvig,et al.  Artificial Intelligence: A Modern Approach , 1995 .

[3]  Drew McDermott,et al.  The 1998 AI Planning Systems Competition , 2000, AI Mag..

[4]  Bernhard Nebel,et al.  The FF Planning System: Fast Plan Generation Through Heuristic Search , 2011, J. Artif. Intell. Res..

[5]  Geoffrey E. Hinton,et al.  Learning Distributed Representations of Concepts Using Linear Relational Embedding , 2001, IEEE Trans. Knowl. Data Eng..

[6]  Malte Helmert,et al.  A Planning Heuristic Based on Causal Graph Analysis , 2004, ICAPS.

[7]  Mariarosaria Taddeo,et al.  Solving the symbol grounding problem: a critical review of fifteen years of research , 2005, J. Exp. Theor. Artif. Intell..

[8]  S. Harnad.  Symbol grounding problem , 1991, Scholarpedia.

[9]  Satoshi Sekine,et al.  A survey of named entity recognition and classification , 2007 .

[10]  Qiang Yang,et al.  Learning action models from plan examples using weighted MAX-SAT , 2007, Artif. Intell..

[11]  L. Steels.  The symbol grounding problem has been solved, so what’s next? , 2008 .

[12]  Ah Chung Tsoi,et al.  The Graph Neural Network Model , 2009, IEEE Transactions on Neural Networks.

[13]  Mark Steedman,et al.  Learning STRIPS Operators from Noisy and Incomplete Observations , 2012, UAI.

[14]  Michal Konkol,et al.  Named Entity Recognition , 2012 .

[15]  T. L. McCluskey,et al.  Acquiring planning domain models using LOCM , 2013, The Knowledge Engineering Review.

[16]  Max Welling,et al.  Semi-supervised Learning with Deep Generative Models , 2014, NIPS.

[17]  Alex Graves,et al.  Recurrent Models of Visual Attention , 2014, NIPS.

[18]  Tom Minka,et al.  A* Sampling , 2014, NIPS.

[19]  Imed Zitouni,et al.  Natural Language Processing of Semitic Languages , 2014, Theory and Applications of Natural Language Processing.

[20]  Yoshua Bengio,et al.  Show, Attend and Tell: Neural Image Caption Generation with Visual Attention , 2015, ICML.

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

[22]  Shane Legg,et al.  Human-level control through deep reinforcement learning , 2015, Nature.

[23]  Yoshua Bengio,et al.  Neural Machine Translation by Jointly Learning to Align and Translate , 2014, ICLR.

[24]  Murray Shanahan,et al.  Towards Deep Symbolic Reinforcement Learning , 2016, ArXiv.

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

[26]  Sergio Gomez Colmenarejo,et al.  Hybrid computing using a neural network with dynamic external memory , 2016, Nature.

[27]  Ali Farhadi,et al.  You Only Look Once: Unified, Real-Time Object Detection , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

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

[29]  Ben Poole,et al.  Categorical Reparameterization with Gumbel-Softmax , 2016, ICLR.

[30]  Tim Rocktäschel,et al.  End-to-end Differentiable Proving , 2017, NIPS.

[31]  Felipe Meneguzzi,et al.  Goal Recognition in Latent Space , 2018, 2018 International Joint Conference on Neural Networks (IJCNN).

[32]  Razvan Pascanu,et al.  Relational inductive biases, deep learning, and graph networks , 2018, ArXiv.

[33]  Masataro Asai.  Photo-Realistic Blocksworld Dataset , 2018, ArXiv.

[34]  Jürgen Schmidhuber,et al.  Relational Neural Expectation Maximization: Unsupervised Discovery of Objects and their Interactions , 2018, ICLR.

[35]  Razvan Pascanu,et al.  Relational Deep Reinforcement Learning , 2018, ArXiv.

[36]  Sebastian Riedel,et al.  Logical Rule Induction and Theory Learning Using Neural Theorem Proving , 2018, ArXiv.

[37]  Alex S. Fukunaga,et al.  Classical Planning in Deep Latent Space: Bridging the Subsymbolic-Symbolic Boundary , 2017, AAAI.

[38]  Steven Schockaert,et al.  Lifted Relational Neural Networks: Efficient Learning of Latent Relational Structures , 2018, J. Artif. Intell. Res..

[39]  Luc De Raedt,et al.  DeepProbLog: Neural Probabilistic Logic Programming , 2018, BNAIC/BENELEARN.