Invariance encoding in sliced-Wasserstein space for image classification with limited training data

Deep convolutional neural networks (CNNs) are broadly considered to be state-of-the-art generic end-to-end image classification systems. However, they are known to underperform when training data are limited and thus require data augmentation strategies that render the method computationally expensive and not always effective. Rather than using a data augmentation strategy to encode invariances as typically done in machine learning, here we propose to mathematically augment a nearest subspace classification model in sliced-Wasserstein space by exploiting certain mathematical properties of the Radon Cumulative Distribution Transform (R-CDT), a recently introduced image transform. We demonstrate that for a particular type of learning problem, our mathematical solution has advantages over data augmentation with deep CNNs in terms of classification accuracy and computational complexity, and is particularly effective under a limited training data setting. The method is simple, effective, computationally efficient, noniterative, and requires no parameters to be tuned. Python code implementing our method is available at [1]. Our method is integrated as a part of the software package PyTransKit, which is available at [2].

[1]  Andrew Zisserman,et al.  Very Deep Convolutional Networks for Large-Scale Image Recognition , 2014, ICLR.

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

[3]  Gustavo K. Rohde,et al.  Cell Image Classification: A Comparative Overview , 2020, Cytometry. Part A : the journal of the International Society for Analytical Cytology.

[4]  Vladimir L. Arlazarov,et al.  ANA HEp-2 cells image classification using number, size, shape and localization of targeted cell regions , 2014, Pattern Recognit..

[5]  Chloe Hutton,et al.  Classification of amyloid status using machine learning with histograms of oriented 3D gradients , 2016, NeuroImage: Clinical.

[6]  Gustavo K. Rohde,et al.  The Radon Cumulative Distribution Transform and Its Application to Image Classification , 2015, IEEE Transactions on Image Processing.

[7]  Sergey Ioffe,et al.  Rethinking the Inception Architecture for Computer Vision , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[8]  Geoffrey E. Hinton,et al.  ImageNet classification with deep convolutional neural networks , 2012, Commun. ACM.

[9]  Tsuyoshi Murata,et al.  {m , 1934, ACML.

[10]  Gustavo K. Rohde,et al.  Parametric Signal Estimation Using the Cumulative Distribution Transform , 2020, IEEE Transactions on Signal Processing.

[11]  Amit K. Roy-Chowdhury,et al.  CNN based region proposals for efficient object detection , 2016, 2016 IEEE International Conference on Image Processing (ICIP).

[12]  Luis Perez,et al.  The Effectiveness of Data Augmentation in Image Classification using Deep Learning , 2017, ArXiv.

[13]  Andreas Holzinger,et al.  Biomedical image augmentation using Augmentor , 2019, Bioinform..

[14]  Gustavo K. Rohde,et al.  Training classifiers with limited data using the Radon cumulative distribution transform , 2018, Medical Imaging.

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

[16]  James E. Fowler,et al.  Nearest Regularized Subspace for Hyperspectral Classification , 2014, IEEE Transactions on Geoscience and Remote Sensing.

[17]  Soheil Kolouri,et al.  Detecting and visualizing cell phenotype differences from microscopy images using transport-based morphometry , 2014, Proceedings of the National Academy of Sciences.

[18]  Ting Liu,et al.  Recent advances in convolutional neural networks , 2015, Pattern Recognit..

[19]  R. Fisher THE USE OF MULTIPLE MEASUREMENTS IN TAXONOMIC PROBLEMS , 1936 .

[20]  P. Alam ‘N’ , 2021, Composites Engineering: An A–Z Guide.

[21]  Jian Sun,et al.  Deep Residual Learning for Image Recognition , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[22]  Robert F. Murphy,et al.  Towards a Systematics for Protein Subcellular Location: Quantitative Description of Protein Localization Patterns and Automated Analysis of Fluorescence Microscope Images , 2000, ISMB.

[23]  Kazuhiro Fukui,et al.  HEp-2 cell classification using rotation invariant co-occurrence among local binary patterns , 2014, Pattern Recognit..

[24]  Mark D. McDonnell,et al.  Understanding Data Augmentation for Classification: When to Warp? , 2016, 2016 International Conference on Digital Image Computing: Techniques and Applications (DICTA).

[25]  M L Mendelsohn,et al.  THE ANALYSIS OF CELL IMAGES * , 1966, Annals of the New York Academy of Sciences.

[26]  Gustavo K. Rohde,et al.  The Cumulative Distribution Transform and Linear Pattern Classification , 2015, Applied and Computational Harmonic Analysis.

[27]  Budiman Minasny,et al.  Using deep learning to predict soil properties from regional spectral data , 2019, Geoderma Regional.

[28]  C. Stolk The Radon transform , 2014 .

[29]  Geoff S. Nitschke,et al.  Improving Deep Learning with Generic Data Augmentation , 2018, 2018 IEEE Symposium Series on Computational Intelligence (SSCI).

[30]  Akram Aldroubi,et al.  Radon cumulative distribution transform subspace modeling for image classification , 2020, ArXiv.

[31]  G. G. Stokes "J." , 1890, The New Yale Book of Quotations.

[32]  Shuzhi Sam Ge,et al.  $k$-NS: A Classifier by the Distance to the Nearest Subspace , 2011, IEEE Transactions on Neural Networks.

[33]  Javad Abbasi Aghamaleki,et al.  Multi-stream CNN for facial expression recognition in limited training data , 2019, Multimedia Tools and Applications.

[34]  Oliver Kramer,et al.  K-Nearest Neighbors , 2013 .

[35]  Yoshua Bengio,et al.  Gradient-based learning applied to document recognition , 1998, Proc. IEEE.

[36]  P. Strand,et al.  2010 18th Euromicro Conference on Parallel, Distributed and Network-based Processing , 2010 .

[37]  Arthur F. Kramer,et al.  Discovery and visualization of structural biomarkers from MRI using transport-based morphometry , 2017, NeuroImage.

[38]  Anne E Carpenter,et al.  Opportunities and obstacles for deep learning in biology and medicine , 2017, bioRxiv.

[39]  Shadpour Demehri,et al.  Enabling early detection of osteoarthritis from presymptomatic cartilage texture maps via transport-based learning , 2020, Proceedings of the National Academy of Sciences.

[40]  Giles M. Foody,et al.  Evaluation of SVM, RVM and SMLR for Accurate Image Classification With Limited Ground Data , 2012, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing.

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

[42]  Dumitru Erhan,et al.  Going deeper with convolutions , 2014, 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[43]  Michał Grochowski,et al.  Data augmentation for improving deep learning in image classification problem , 2018, 2018 International Interdisciplinary PhD Workshop (IIPhDW).

[44]  E. T. Quinto,et al.  An Introduction to X-ray tomography and Radon Transforms , 2006 .

[45]  Supratik Mukhopadhyay,et al.  Unsupervised Learning using Pretrained CNN and Associative Memory Bank , 2018, 2018 International Joint Conference on Neural Networks (IJCNN).

[46]  Haipeng Wang,et al.  Application of deep-learning algorithms to MSTAR data , 2015, 2015 IEEE International Geoscience and Remote Sensing Symposium (IGARSS).

[47]  Natalia Gimelshein,et al.  PyTorch: An Imperative Style, High-Performance Deep Learning Library , 2019, NeurIPS.

[48]  Kilian Q. Weinberger,et al.  Deep Networks with Stochastic Depth , 2016, ECCV.

[49]  Gorjan Alagic,et al.  #p , 2019, Quantum information & computation.

[50]  P. Alam ‘G’ , 2021, Composites Engineering: An A–Z Guide.

[51]  Jaeyeon Lee,et al.  ETRI-Activity3D: A Large-Scale RGB-D Dataset for Robots to Recognize Daily Activities of the Elderly , 2020, 2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS).

[52]  Robert F. Murphy,et al.  A neural network classifier capable of recognizing the patterns of all major subcellular structures in fluorescence microscope images of HeLa cells , 2001, Bioinform..

[53]  Kui Ren,et al.  Adversarial Attacks and Defenses in Deep Learning , 2020, Engineering.

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

[55]  Patrick Neary Automatic Hyperparameter Tuning in Deep Convolutional Neural Networks Using Asynchronous Reinforcement Learning , 2018, 2018 IEEE International Conference on Cognitive Computing (ICCC).

[57]  Yair Weiss,et al.  Why do deep convolutional networks generalize so poorly to small image transformations? , 2018, J. Mach. Learn. Res..

[58]  Chi-Hua Chen,et al.  Hyperparameter Tuning Deep Learning for Diabetic Retinopathy Fundus Image Classification , 2020, IEEE Access.

[59]  Jana Kosecka,et al.  3D Bounding Box Estimation Using Deep Learning and Geometry , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[60]  Taghi M. Khoshgoftaar,et al.  A survey on Image Data Augmentation for Deep Learning , 2019, Journal of Big Data.

[61]  Fan Zhang,et al.  Low-cost and high-performance data augmentation for deep-learning-based skin lesion classification , 2021, ArXiv.

[62]  Kibok Lee,et al.  A Simple Unified Framework for Detecting Out-of-Distribution Samples and Adversarial Attacks , 2018, NeurIPS.

[63]  Lior Shamir,et al.  WND-CHARM: Multi-purpose image classification using compound image transforms , 2008, Pattern Recognit. Lett..

[64]  Kyandoghere Kyamakya,et al.  CNN based high performance computing for real time image processing on GPU , 2011, Proceedings of the Joint INDS'11 & ISTET'11.

[65]  Gustavo K. Rohde,et al.  Nearest Subspace Search in The Signed Cumulative Distribution Transform Space for 1D Signal Classification , 2021 .

[66]  Daniel L. Rubin,et al.  Differential Data Augmentation Techniques for Medical Imaging Classification Tasks , 2017, AMIA.

[67]  Lina Yao,et al.  Deep Neural Network Hyperparameter Optimization with Orthogonal Array Tuning , 2019, ICONIP.

[68]  Ajmal Mian,et al.  Threat of Adversarial Attacks on Deep Learning in Computer Vision: A Survey , 2018, IEEE Access.

[69]  F. Natterer The Mathematics of Computerized Tomography , 1986 .

[70]  Lu Tian,et al.  A High-Performance CNN Processor Based on FPGA for MobileNets , 2019, 2019 29th International Conference on Field Programmable Logic and Applications (FPL).

[71]  Kevin McGuinness,et al.  Deep Learning for Computer Vision IMVIP 2019 : Irish Machine Vision and ImageProcessing 2019 Comparing Data Augmentation Strategies for Deep Image Classification , 2019 .

[72]  Dong Yu,et al.  Deep Learning: Methods and Applications , 2014, Found. Trends Signal Process..

[73]  Patrice Y. Simard,et al.  Best practices for convolutional neural networks applied to visual document analysis , 2003, Seventh International Conference on Document Analysis and Recognition, 2003. Proceedings..

[74]  Roger C. Tam,et al.  Manifold Learning of Brain MRIs by Deep Learning , 2013, MICCAI.

[75]  P. Alam ‘A’ , 2021, Composites Engineering: An A–Z Guide.

[76]  Hunter Blanton,et al.  Inconsistent Performance of Deep Learning Models on Mammogram Classification. , 2020, Journal of the American College of Radiology : JACR.

[77]  P. Alam ‘L’ , 2021, Composites Engineering: An A–Z Guide.