Scientific Discovery by Generating Counterfactuals using Image Translation

Model explanation techniques play a critical role in understanding the source of a model’s performance and making its decisions transparent. Here we investigate if explanation techniques can also be used as a mechanism for scientific discovery. We make three contributions: first, we propose a framework to convert predictions from explanation techniques to a mechanism of discovery. Second, we show how generative models in combination with black-box predictors can be used to generate hypotheses (without human priors) that can be critically examined. Third, with these techniques we study classification models for retinal images predicting Diabetic Macular Edema (DME), where recent work [30] showed that a CNN trained on these images is likely learning novel features in the image. We demonstrate that the proposed framework is able to explain the underlying scientific mechanism, thus bridging the gap between the model’s performance and human understanding.

[1]  Mark Sandler,et al.  CycleGAN, a Master of Steganography , 2017, ArXiv.

[2]  Thomas Brox,et al.  Striving for Simplicity: The All Convolutional Net , 2014, ICLR.

[3]  Michael V. McConnell,et al.  Prediction of cardiovascular risk factors from retinal fundus photographs via deep learning , 2017, Nature Biomedical Engineering.

[4]  Rob Fergus,et al.  Visualizing and Understanding Convolutional Networks , 2013, ECCV.

[5]  Catherine Egan,et al.  SDOCT Imaging to Identify Macular Pathology in Patients Diagnosed with Diabetic Maculopathy by a Digital Photographic Retinal Screening Programme , 2011, PloS one.

[6]  Ramprasaath R. Selvaraju,et al.  Grad-CAM: Visual Explanations from Deep Networks via Gradient-Based Localization , 2017, 2017 IEEE International Conference on Computer Vision (ICCV).

[7]  Brian Pollack,et al.  Explanation by Progressive Exaggeration , 2020, ICLR.

[8]  T. Wong,et al.  Epidemiology of diabetic retinopathy, diabetic macular edema and related vision loss , 2015, Eye and Vision.

[9]  Andrea Vedaldi,et al.  Understanding deep image representations by inverting them , 2014, 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[10]  Carlos Guestrin,et al.  "Why Should I Trust You?": Explaining the Predictions of Any Classifier , 2016, ArXiv.

[11]  Yoshua Bengio,et al.  Generative Adversarial Nets , 2014, NIPS.

[12]  edited by Paul Riordan-Eva and John P. Whitcher.  Vaughan & Asbury's General Ophthalmology , 2003 .

[13]  Subhashini Venugopalan,et al.  Development and Validation of a Deep Learning Algorithm for Detection of Diabetic Retinopathy in Retinal Fundus Photographs. , 2016, JAMA.

[14]  S. Harding,et al.  Sensitivity and specificity of photography and direct ophthalmoscopy in screening for sight threatening eye disease: the Liverpool diabetic eye study , 1995, BMJ.

[15]  Andrea Vedaldi,et al.  Interpretable Explanations of Black Boxes by Meaningful Perturbation , 2017, 2017 IEEE International Conference on Computer Vision (ICCV).

[16]  Bolei Zhou,et al.  Network Dissection: Quantifying Interpretability of Deep Visual Representations , 2017, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[17]  Amit Dhurandhar,et al.  Explanations based on the Missing: Towards Contrastive Explanations with Pertinent Negatives , 2018, NeurIPS.

[18]  John P. Cunningham,et al.  Discriminative Regularization for Latent Variable Models with Applications to Electrocardiography , 2019, ICML.

[19]  Jr-hau He,et al.  Spontaneous solar water splitting with decoupling of light absorption and electrocatalysis using silicon back-buried junction , 2020, Nature Communications.

[20]  Neil M Bressler,et al.  Comparison of Prevalence of Diabetic Macular Edema Based on Monocular Fundus Photography vs Optical Coherence Tomography. , 2016, JAMA ophthalmology.

[21]  Jonathan Krause,et al.  Grader variability and the importance of reference standards for evaluating machine learning models for diabetic retinopathy , 2017, Ophthalmology.

[22]  Alexei A. Efros,et al.  Unpaired Image-to-Image Translation Using Cycle-Consistent Adversarial Networks , 2017, 2017 IEEE International Conference on Computer Vision (ICCV).

[23]  Kate Saenko,et al.  RISE: Randomized Input Sampling for Explanation of Black-box Models , 2018, BMVC.

[24]  Andrea Vedaldi,et al.  Understanding Deep Networks via Extremal Perturbations and Smooth Masks , 2019, 2019 IEEE/CVF International Conference on Computer Vision (ICCV).

[25]  Subhashini Venugopalan,et al.  Predicting optical coherence tomography-derived diabetic macular edema grades from fundus photographs using deep learning , 2018, Nature Communications.

[26]  Oluwasanmi Koyejo,et al.  Towards Realistic Individual Recourse and Actionable Explanations in Black-Box Decision Making Systems , 2019, ArXiv.

[27]  Yong Han,et al.  Generative Counterfactual Introspection for Explainable Deep Learning , 2019, 2019 IEEE Global Conference on Signal and Information Processing (GlobalSIP).

[28]  Aaron C. Courville,et al.  Improved Training of Wasserstein GANs , 2017, NIPS.

[29]  David Duvenaud,et al.  Explaining Image Classifiers by Counterfactual Generation , 2018, ICLR.

[30]  Ankur Taly,et al.  Axiomatic Attribution for Deep Networks , 2017, ICML.

[31]  Martin Wattenberg,et al.  SmoothGrad: removing noise by adding noise , 2017, ArXiv.

[32]  Pouya Samangouei,et al.  ExplainGAN: Model Explanation via Decision Boundary Crossing Transformations , 2018, ECCV.

[33]  Tolga Bolukbasi,et al.  XRAI: Better Attributions Through Regions , 2019, 2019 IEEE/CVF International Conference on Computer Vision (ICCV).