• Corpus ID: 215413641

Extending Unsupervised Neural Image Compression With Supervised Multitask Learning

  title={Extending Unsupervised Neural Image Compression With Supervised Multitask Learning},
  author={David Tellez and Diederik J. H{\"o}ppener and Cornelis Verhoef and Dirk J. Gr{\"u}nhagen and Pieter Nierop and Michal Drozdzal and Jeroen van der Laak and Francesco Ciompi},
We focus on the problem of training convolutional neural networks on gigapixel histopathology images to predict image-level targets. For this purpose, we extend Neural Image Compression (NIC), an image compression framework that reduces the dimensionality of these images using an encoder network trained unsupervisedly. We propose to train this encoder using supervised multitask learning (MTL) instead. We applied the proposed MTL NIC to two histopathology datasets and three tasks. First, we… 
Neural image compression for non-small cell lung cancer subtype classification in H&E stained whole-slide images
Classification of non-small-cell lung cancer (NSCLC) into adenocarcinoma (LUAD) and squamous cell carcinoma (LUSC) via histopathology is a vital prerequisite to select the appropriate treatment for
Evaluating Generic Auto-ML Tools for Computational Pathology
It is found that the default CNN architectures and parameterizations of the evaluated AutoML tools already yielded classi cation performance on par with the original publications, which would allow developers of software solutions for computational pathology to focus on harder-to-automate tasks such as data curation.
Few-shot weakly supervised detection and retrieval in histopathology whole-slide images
In this work, we propose a deep learning system for weakly supervised object detection in digital pathology whole slide images. We designed the system to be organ- and object-agnostic, and to be
Ranking Micro-Influencers: a Novel Multi-Task Learning and Interpretable Framework
A novel multi-task learning framework to improve the state of the art in micro-influencer ranking based on multimedia content and provides an effective visual method for interpreting the authors' models’ decisions, which can also be used to inform brands’ media strategies.
DeepSMILE: Self-supervised heterogeneity-aware multiple instance learning for DNA damage response defect classification directly from H&E whole-slide images
We propose a Deep learning-based weak label learning method for analysing whole slide images (WSIs) of Hematoxylin and Eosin (H&E) stained tumorcells not requiring pixel-level or tile-level
Multiple Instance Captioning: Learning Representations from Histopathology Textbooks and Articles
It is shown that ARCH is the only CP dataset to (ARCH-)rival its computer vision analog MS-COCO Captions, and conjecture that an encoder pre-trained on dense image captions learns transferable representations for most CP tasks.


Neural Image Compression for Gigapixel Histopathology Image Analysis
The proposed Neural Image Compression (NIC), a two-step method to build convolutional neural networks for gigapixel image analysis solely using weak image-level labels, can exploit visual cues associated with image- level labels successfully, integrating both global and local visual information.
Predicting Cancer with a Recurrent Visual Attention Model for Histopathology Images
A novel attention-based model is presented that is capable of attending to the most discriminative regions of an image by adaptively selecting a limited sequence of locations and only processing the selected areas of tissues.
Needles in Haystacks: On Classifying Tiny Objects in Large Images
There exists a limit to signal-to-noise below which CNNs fail to generalize and that this limit is affected by dataset size - more data leading to better performances; however, in general, higher capacity models exhibit better generalization.
Weakly Supervised Learning for Whole Slide Lung Cancer Image Classification
A weakly supervised approach for fast and effective classification on whole slide lung cancer images that takes advantage of a patch-based fully convolutional network for discriminative block retrieval and context-aware feature selection and aggregation strategies are proposed to generate globally holistic WSI descriptor.
Quantifying the effects of data augmentation and stain color normalization in convolutional neural networks for computational pathology
A novel unsupervised method to perform stain colornormalization using a neural network is proposed, providing practical guidelines on how to use stain color augmentation and stain color normalization in future computational pathology applications.
Reinforced Auto-Zoom Net: Towards Accurate and Fast Breast Cancer Segmentation in Whole-slide Images
This paper proposes a simple yet efficient framework Reinforced Auto-Zoom Net (RAZN), motivated by the zoom-in operation of a pathologist using a digital microscope, which outperforms both single-scale and multi-scale baseline approaches, achieving better accuracy at low inference cost.
Patch-Based Convolutional Neural Network for Whole Slide Tissue Image Classification
A novel Expectation-Maximization (EM) based method is formulated that automatically locates discriminative patches robustly by utilizing the spatial relationships of patches and applies it to the classification of glioma and non-small-cell lung carcinoma cases into subtypes.
Siamese Neural Networks for One-Shot Image Recognition
A method for learning siamese neural networks which employ a unique structure to naturally rank similarity between inputs and is able to achieve strong results which exceed those of other deep learning models with near state-of-the-art performance on one-shot classification tasks.
Large scale tissue histopathology image classification, segmentation, and visualization via deep convolutional activation features
The framework proposed is a simple, efficient and effective system for histopathology image automatic analysis that successfully transfer ImageNet knowledge as deep convolutional activation features to the classification and segmentation of histopathological images with little training data.
Attention-based Deep Multiple Instance Learning
This paper proposes a neural network-based permutation-invariant aggregation operator that corresponds to the attention mechanism that achieves comparable performance to the best MIL methods on benchmark MIL datasets and outperforms other methods on a MNIST-based MIL dataset and two real-life histopathology datasets without sacrificing interpretability.