Interpretable Actions: Controlling Experts with Understandable Commands

Despite the prevalence of deep neural networks, their single most cited drawback is that, even when successful, their operations are inscrutable. For many applications, the desired outputs are the composition of externally-defined bases. For such decomposable domains, we present a two-stage learning procedure producing combinations of the external bases which are trivially extractable from the network. In the first stage, the set of external bases that will form the solution are modeled as differentiable generator modules, controlled by the same parameters as the external bases. In the second stage, a controller network is created that selects parameters for those generators, either successively or in parallel, to compose the final solution. Through three tasks, we concretely demonstrate how our system yields readily understandable commands. In one, we introduce a new form of artistic style transfer, learning to draw and color with crayons, in which the transformation of a photograph or painting occurs not as a single monolithic computation, but by the composition of thousands of individual, visualizable strokes. The other two tasks, single-pass function approximation with arbitrary bases and shape-based synthesis, show how our approach produces understandable and extractable actions in two disparate domains.

[1]  Geoffrey E. Hinton,et al.  Symbols Among the Neurons: Details of a Connectionist Inference Architecture , 1985, IJCAI.

[2]  Wenbin Cai,et al.  A Unified Framework for Generalizable Style Transfer: Style and Content Separation , 2018, IEEE Transactions on Image Processing.

[3]  Lucas Theis,et al.  Lossy Image Compression with Compressive Autoencoders , 2017, ICLR.

[4]  Leslie Pack Kaelbling,et al.  Modular meta-learning , 2018, CoRL.

[5]  Armando Solar-Lezama,et al.  Learning to Infer Graphics Programs from Hand-Drawn Images , 2017, NeurIPS.

[6]  Björn Ommer,et al.  Content and Style Disentanglement for Artistic Style Transfer , 2019, 2019 IEEE/CVF International Conference on Computer Vision (ICCV).

[7]  M. Kramer Nonlinear principal component analysis using autoassociative neural networks , 1991 .

[8]  Geoffrey E. Hinton,et al.  Transforming Auto-Encoders , 2011, ICANN.

[9]  Li Fei-Fei,et al.  ImageNet: A large-scale hierarchical image database , 2009, CVPR.

[10]  Michael J. Black,et al.  OpenDR: An Approximate Differentiable Renderer , 2014, ECCV.

[11]  Leon A. Gatys,et al.  A Neural Algorithm of Artistic Style , 2015, ArXiv.

[12]  Geoffrey E. Hinton,et al.  Dynamic Routing Between Capsules , 2017, NIPS.

[13]  Sebastian Nowozin,et al.  The informed sampler: A discriminative approach to Bayesian inference in generative computer vision models , 2014, Comput. Vis. Image Underst..

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

[15]  Li Fei-Fei,et al.  Perceptual Losses for Real-Time Style Transfer and Super-Resolution , 2016, ECCV.

[16]  Pushmeet Kohli,et al.  Vision-as-Inverse-Graphics: Obtaining a Rich 3D Explanation of a Scene from a Single Image , 2017, 2017 IEEE International Conference on Computer Vision Workshops (ICCVW).

[17]  Hod Lipson,et al.  Understanding Neural Networks Through Deep Visualization , 2015, ArXiv.

[18]  Bruce G. Baumgart,et al.  Geometric modeling for computer vision. , 1974 .

[19]  Fujio Yamaguchi,et al.  Computer-Aided Geometric Design , 2002, Springer Japan.

[20]  Oriol Vinyals,et al.  Synthesizing Programs for Images using Reinforced Adversarial Learning , 2018, ICML.

[21]  Ole Winther,et al.  Autoencoding beyond pixels using a learned similarity metric , 2015, ICML.

[22]  Paul Smolensky Neural and Conceptual Interpretations of Parallel Distributed Processing Models , 1986 .

[23]  Alun D. Preece,et al.  Interpretability of deep learning models: A survey of results , 2017, 2017 IEEE SmartWorld, Ubiquitous Intelligence & Computing, Advanced & Trusted Computed, Scalable Computing & Communications, Cloud & Big Data Computing, Internet of People and Smart City Innovation (SmartWorld/SCALCOM/UIC/ATC/CBDCom/IOP/SCI).

[24]  Stéphane Mallat,et al.  Matching pursuits with time-frequency dictionaries , 1993, IEEE Trans. Signal Process..

[25]  Joshua B. Tenenbaum,et al.  Automatic Construction and Natural-Language Description of Nonparametric Regression Models , 2014, AAAI.

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

[27]  Jiajun Wu,et al.  Learning to Describe Scenes with Programs , 2018, ICLR.

[28]  Jiajun Wu,et al.  Neural Scene De-rendering , 2017, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[29]  Douglas Eck,et al.  A Neural Representation of Sketch Drawings , 2017, ICLR.

[30]  Dan Boneh,et al.  Ensemble Adversarial Training: Attacks and Defenses , 2017, ICLR.

[31]  Tijmen Tieleman,et al.  Optimizing Neural Networks that Generate Iimages , 2014 .

[32]  Gustavo Patow,et al.  A Survey of Inverse Rendering Problems , 2003, Comput. Graph. Forum.

[33]  Subhransu Maji,et al.  Neural Shape Parsers for Constructive Solid Geometry , 2019, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[34]  Cynthia Rudin,et al.  Deep Learning for Case-based Reasoning through Prototypes: A Neural Network that Explains its Predictions , 2017, AAAI.

[35]  Andrew Zisserman,et al.  Spatial Transformer Networks , 2015, NIPS.

[36]  Michael I. Jordan,et al.  Learning piecewise control strategies in a modular neural network architecture , 1993, IEEE Trans. Syst. Man Cybern..

[37]  Romain Hennequin,et al.  Making Neural Networks Interpretable with Attribution: Application to Implicit Signals Prediction , 2020, RecSys.

[38]  Joshua B. Tenenbaum,et al.  Inverse Graphics with Probabilistic CAD Models , 2014, ArXiv.

[39]  Ersin Yumer,et al.  ST-GAN: Spatial Transformer Generative Adversarial Networks for Image Compositing , 2018, 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.

[40]  J. Jiang,et al.  Image compression with neural networks - A survey , 1999, Signal Process. Image Commun..

[41]  Joshua B. Tenenbaum,et al.  Exploiting compositionality to explore a large space of model structures , 2012, UAI.

[42]  Alexei A. Efros,et al.  Image-to-Image Translation with Conditional Adversarial Networks , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[43]  Jiajun Wu,et al.  Learning to Infer and Execute 3D Shape Programs , 2019, ICLR.

[44]  Sergey Levine,et al.  Learning modular neural network policies for multi-task and multi-robot transfer , 2016, 2017 IEEE International Conference on Robotics and Automation (ICRA).

[45]  Ge Wang,et al.  On Interpretability of Artificial Neural Networks , 2020, ArXiv.

[46]  Teresa B. Ludermir Extracting Rules From Neural Networks: AData Mining Approach , 1970 .

[47]  Andrea Vedaldi,et al.  Visualizing Deep Convolutional Neural Networks Using Natural Pre-images , 2015, International Journal of Computer Vision.

[48]  Simon Lucey,et al.  Inverse Compositional Spatial Transformer Networks , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[49]  Susumu Tachi,et al.  A modular neural network architecture for inverse kinematics model learning , 2001, Neurocomputing.

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

[51]  Armando Solar-Lezama,et al.  Write, Execute, Assess: Program Synthesis with a REPL , 2019, NeurIPS.

[52]  Hang Su,et al.  Towards Interpretable Deep Neural Networks by Leveraging Adversarial Examples , 2017, ArXiv.

[53]  Abhishek Das,et al.  Grad-CAM: Visual Explanations from Deep Networks via Gradient-Based Localization , 2016, 2017 IEEE International Conference on Computer Vision (ICCV).

[54]  Quanshi Zhang,et al.  Interpretable Convolutional Neural Networks , 2017, 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.

[55]  Wojciech Samek,et al.  Methods for interpreting and understanding deep neural networks , 2017, Digit. Signal Process..

[56]  Parul Parashar,et al.  Neural Networks in Machine Learning , 2014 .

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

[58]  Premkumar Natarajan,et al.  CapsuleGAN: Generative Adversarial Capsule Network , 2018, ECCV Workshops.

[59]  Dan Klein,et al.  Deep Compositional Question Answering with Neural Module Networks , 2015, ArXiv.