Chromosome classification is critical for karyotyping in abnormality diagnosis. To expedite the diagnosis, we present a novel method named Varifocal-Net for simultaneous classification of chromosome's type and polarity using deep convolutional networks. The approach consists of one global-scale network (G-Net) and one local-scale network (L-Net). It follows three stages. The first stage is to learn both global and local features. We extract global features and detect finer local regions via the G-Net. By proposing a varifocal mechanism, we zoom into local parts and extract local features via the L-Net. Residual learning and multi-task learning strategies are utilized to promote high-level feature extraction. The detection of discriminative local parts is fulfilled by a localization subnet of the G-Net, whose training process involves both supervised and weakly-supervised learning. The second stage is to build two multi-layer perceptron classifiers that exploit features of both two scales to boost classification performance. The third stage is to introduce a dispatch strategy of assigning each chromosome to a type within each patient case, by utilizing the domain knowledge of karyotyping. Evaluation results from 1909 karyotyping cases showed that the proposed Varifocal-Net achieved the highest accuracy per patient case (%) 99.2 for both type and polarity tasks. It outperformed state-of-the-art methods, demonstrating the effectiveness of our varifocal mechanism, multi-scale feature ensemble, and dispatch strategy. The proposed method has been applied to assist practical karyotype diagnosis.
Deformable registration has been one of the pillars of biomedical image computing. Conventional approaches refer to the definition of a similarity criterion that, once endowed with a deformation model and a smoothness constraint, determines the optimal transformation to align two given images. The definition of this metric function is among the most critical aspects of the registration process. We argue that incorporating semantic information (in the form of anatomical segmentation maps) into the registration process will further improve the accuracy of the results. In this paper, we propose a novel weakly supervised approach to learn domain specific aggregations of conventional metrics using anatomical segmentations. This combination is learned using latent structured support vector machines (LSSVM). The learned matching criterion is integrated within a metric free optimization framework based on graphical models, resulting in a multi-metric algorithm endowed with a spatially varying similarity metric function conditioned on the anatomical structures. We provide extensive evaluation on three different datasets of CT and MRI images, showing that learned multi-metric registration outperforms single-metric approaches based on conventional similarity measures.
Performing data augmentation for learning deep neural networks is known to be important for training visual recognition systems. By artificially increasing the number of training examples, it helps reducing overfitting and improves generalization. While simple image transformations can already improve predictive performance in most vision tasks, larger gains can be obtained by leveraging task-specific prior knowledge. In this work, we consider object detection, semantic and instance segmentation and augment the training images by blending objects in existing scenes, using instance segmentation annotations. We observe that randomly pasting objects on images hurts the performance, unless the object is placed in the right context. To resolve this issue, we propose an explicit context model by using a convolutional neural network, which predicts whether an image region is suitable for placing a given object or not. In our experiments, we show that our approach is able to improve object detection, semantic and instance segmentation on the PASCAL VOC12 and COCO datasets, with significant gains in a limited annotation scenario, i.e. when only one category is annotated. We also show that the method is not limited to datasets that come with expensive pixel-wise instance annotations and can be used when only bounding boxes are available, by employing weakly-supervised learning for instance masks approximation.
A major element of depth perception and 3D understanding is the ability to predict the 3D layout of a scene and its contained objects for a novel pose. Indoor environments are particularly suitable for novel view prediction, since the set of objects in such environments is relatively restricted. In this work we address the task of 3D prediction especially for indoor scenes by leveraging only weak supervision. In the literature 3D scene prediction is usually solved via a 3D voxel grid. However, such methods are limited to estimating rather coarse 3D voxel grids, since predicting entire voxel spaces has large computational costs. Hence, our method operates in image-space rather than in voxel space, and the task of 3D estimation essentially becomes a depth image completion problem. We propose a novel approach to easily generate training data containing depth maps with realistic occlusions, and subsequently train a network for completing those occluded regions. Using multiple publicly available dataset~\cite{song2017semantic,Silberman:ECCV12} we benchmark our method against existing approaches and are able to obtain superior performance. We further demonstrate the flexibility of our method by presenting results for new view synthesis of RGB-D images.
We present a simple yet efficient approach capable of training deep neural networks on large-scale weakly-supervised web images, which are crawled raw from the Internet by using text queries, without any human annotation. We develop a principled learning strategy by leveraging curriculum learning, with the goal of handling a massive amount of noisy labels and data imbalance effectively. We design a new learning curriculum by measuring the complexity of data using its distribution density in a feature space, and rank the complexity in an unsupervised manner. This allows for an efficient implementation of curriculum learning on large-scale web images, resulting in a high-performance CNN model, where the negative impact of noisy labels is reduced substantially. Importantly, we show by experiments that those images with highly noisy labels can surprisingly improve the generalization capability of the model, by serving as a manner of regularization. Our approaches obtain state-of-the-art performance on four benchmarks: WebVision, ImageNet, Clothing-1M and Food-101. With an ensemble of multiple models, we achieved a top-5 error rate of 5.2% on the WebVision challenge for 1000-category classification. This result was the top performance by a wide margin, outperforming second place by a nearly 50% relative error rate. Code and models are available at: https://github.com/MalongTech/CurriculumNet .
We propose a method to perform audio event detection under the common constraint that only limited training data are available. In training a deep learning system to perform audio event detection, two practical problems arise. Firstly, most datasets are "weakly labelled" having only a list of events present in each recording without any temporal information for training. Secondly, deep neural networks need a very large amount of labelled training data to achieve good quality performance, yet in practice it is difficult to collect enough samples for most classes of interest. In this paper, we propose a data-efficient training of a stacked convolutional and recurrent neural network. This neural network is trained in a multi instance learning setting for which we introduce a new loss function that leads to improved training compared to the usual approaches for weakly supervised learning. We successfully test our approach on two low-resource datasets that lack temporal labels.
Computer-aided techniques may lead to more accurate and more acces-sible diagnosis of thorax diseases on chest radiography. Despite the success of deep learning-based solutions, this task remains a major challenge in smart healthcare, since it is intrinsically a weakly supervised learning problem. In this paper, we incorporate the attention mechanism into a deep convolutional neural network, and thus propose the ChestNet model to address effective diagnosis of thorax diseases on chest radiography. This model consists of two branches: a classification branch serves as a uniform feature extraction-classification network to free users from troublesome handcrafted feature extraction, and an attention branch exploits the correlation between class labels and the locations of patholog-ical abnormalities and allows the model to concentrate adaptively on the patholog-ically abnormal regions. We evaluated our model against three state-of-the-art deep learning models on the Chest X-ray 14 dataset using the official patient-wise split. The results indicate that our model outperforms other methods, which use no extra training data, in diagnosing 14 thorax diseases on chest radiography.
We describe a novel weakly supervised deep learning framework that combines both the discriminative and generative models to learn meaningful representation in the multiple instance learning (MIL) setting. MIL is a weakly supervised learning problem where labels are associated with groups of instances (referred as bags) instead of individual instances. To address the essential challenge in MIL problems raised from the uncertainty of positive instances label, we use a discriminative model regularized by variational autoencoders (VAEs) to maximize the differences between latent representations of all instances and negative instances. As a result, the hidden layer of the variational autoencoder learns meaningful representation. This representation can effectively be used for MIL problems as illustrated by better performance on the standard benchmark datasets comparing to the state-of-the-art approaches. More importantly, unlike most related studies, the proposed framework can be easily scaled to large dataset problems, as illustrated by the audio event detection and segmentation task. Visualization also confirms the effectiveness of the latent representation in discriminating positive and negative classes.
Knowledge Matters: Importance of Prior Information for Optimization [7], by Gulcehre et. al., sought to establish the limits of current black-box, deep learning techniques by posing problems which are difficult to learn without engineering knowledge into the model or training procedure. In our work, we completely solve the previous Knowledge Matters problem using a generic model, pose a more difficult and scalable problem, All-Pairs, and advance this new problem by introducing a new learned, spatially-varying histogram model called TypeNet which outperforms conventional models on the problem. We present results on All-Pairs where our model achieves 100% test accuracy while the best ResNet models achieve 79% accuracy. In addition, our model is more than an order of magnitude smaller than Resnet-34. The challenge of solving larger-scale All-Pairs problems with high accuracy is presented to the community for investigation.
We consider a multilingual weakly supervised learning scenario where knowledge from annotated corpora in a resource-rich language is transferred via bitext to guide the learning in other languages. Past approaches project labels across bitext and use them as features or gold labels for training. We propose a new method that projects model expectations rather than labels, which facilities transfer of model uncertainty across language boundaries. We encode expectations as constraints and train a discriminative CRF model using Generalized Expectation Criteria (Mann and McCallum, 2010). Evaluated on standard Chinese-English and German-English NER datasets, our method demonstrates F1 scores of 64% and 60% when no labeled data is used. Attaining the same accuracy with supervised CRFs requires 12k and 1.5k labeled sentences. Furthermore, when combined with labeled examples, our method yields significant improvements over state-of-the-art supervised methods, achieving best reported numbers to date on Chinese OntoNotes and German CoNLL-03 datasets.
Action recognition is an important problem in multimedia understanding. This paper addresses this problem by building an expressive compositional action model. We model one action instance in the video with an ensemble of spatio-temporal compositions: a number of discrete temporal anchor frames, each of which is further decomposed to a layout of deformable parts. In this way, our model can identify a Spatio-Temporal And-Or Graph (STAOG) to represent the latent structure of actions e.g. triple jumping, swinging and high jumping. The STAOG model comprises four layers: (i) a batch of leaf-nodes in bottom for detecting various action parts within video patches; (ii) the or-nodes over bottom, i.e. switch variables to activate their children leaf-nodes for structural variability; (iii) the and-nodes within an anchor frame for verifying spatial composition; and (iv) the root-node at top for aggregating scores over temporal anchor frames. Moreover, the contextual interactions are defined between leaf-nodes in both spatial and temporal domains. For model training, we develop a novel weakly supervised learning algorithm which iteratively determines the structural configuration (e.g. the production of leaf-nodes associated with the or-nodes) along with the optimization of multi-layer parameters. By fully exploiting spatio-temporal compositions and interactions, our approach handles well large intra-class action variance (e.g. different views, individual appearances, spatio-temporal structures). The experimental results on the challenging databases demonstrate superior performance of our approach over other competing methods.
When humans describe images they tend to use combinations of nouns and adjectives, corresponding to objects and their associated attributes respectively. To generate such a description automatically, one needs to model objects, attributes and their associations. Conventional methods require strong annotation of object and attribute locations, making them less scalable. In this paper, we model object-attribute associations from weakly labelled images, such as those widely available on media sharing sites (e.g. Flickr), where only image-level labels (either object or attributes) are given, without their locations and associations. This is achieved by introducing a novel weakly supervised non-parametric Bayesian model. Once learned, given a new image, our model can describe the image, including objects, attributes and their associations, as well as their locations and segmentation. Extensive experiments on benchmark datasets demonstrate that our weakly supervised model performs at par with strongly supervised models on tasks such as image description and retrieval based on object-attribute associations.
In this study, a spectral graph-theoretic grouping strategy for weakly supervised classification is introduced, where a limited number of labelled samples and a larger set of unlabelled samples are used to construct a larger annotated training set composed of strongly labelled and weakly labelled samples. The inherent relationship between the set of strongly labelled samples and the set of unlabelled samples is established via spectral grouping, with the unlabelled samples subsequently weakly annotated based on the strongly labelled samples within the associated spectral groups. A number of similarity graph models for spectral grouping, including two new similarity graph models introduced in this study, are explored to investigate their performance in the context of weakly supervised classification in handling different types of data. Experimental results using benchmark datasets as well as real EMG datasets demonstrate that the proposed approach to weakly supervised classification can provide noticeable improvements in classification performance, and that the proposed similarity graph models can lead to ultimate learning results that are either better than or on a par with existing similarity graph models in the context of spectral grouping for weakly supervised classification.
Recent advances in supervised salient object detection has resulted in significant performance on benchmark datasets. Training such models, however, requires expensive pixel-wise annotations of salient objects. Moreover, many existing salient object detection models assume that at least one salient object exists in the input image. Such an assumption often leads to less appealing saliency maps on the background images, which contain no salient object at all. To avoid the requirement of expensive pixel-wise salient region annotations, in this paper, we study weakly supervised learning approaches for salient object detection. Given a set of background images and salient object images, we propose a solution toward jointly addressing the salient object existence and detection tasks. We adopt the latent SVM framework and formulate the two problems together in a single integrated objective function: saliency labels of superpixels are modeled as hidden variables and involved in a classification term conditioned to the salient object existence variable, which in turn depends on both global image and regional saliency features and saliency label assignment. Experimental results on benchmark datasets validate the effectiveness of our proposed approach.
The scarcity of data annotated at the desired level of granularity is a recurring issue in many applications. Significant amounts of effort have been devoted to developing weakly supervised methods tailored to each individual setting, which are often carefully designed to take advantage of the particular properties of weak supervision regimes, form of available data and prior knowledge of the task at hand. Unfortunately, it is difficult to adapt these methods to new tasks and/or forms of data, which often require different weak supervision regimes or models. We present a general-purpose method that can solve any weakly supervised learning problem irrespective of the weak supervision regime or the model. The proposed method turns any off-the-shelf strongly supervised classifier into a weakly supervised classifier and allows the user to specify any arbitrary weakly supervision regime via a loss function. We apply the method to several different weak supervision regimes and demonstrate competitive results compared to methods specifically engineered for those settings.
Traditional approaches for handwritten Chinese character recognition suffer in classifying similar characters. In this paper, we propose to discriminate similar handwritten Chinese characters by using weakly supervised learning. Our approach learns a discriminative SVM for each similar pair which simultaneously localizes the discriminative region of similar character and makes the classification. For the first time, similar handwritten Chinese character recognition (SHCCR) is formulated as an optimization problem extended from SVM. We also propose a novel feature descriptor, Gradient Context, and apply bag-of-words model to represent regions with different scales. In our method, we do not need to select a sized-fixed sub-window to differentiate similar characters. The unconstrained property makes our method well adapted to high variance in the size and position of discriminative regions in similar handwritten Chinese characters. We evaluate our proposed approach over the CASIA Chinese character data set and the results show that our method outperforms the state of the art.
Recent work on scene classification still makes use of generic CNN features in a rudimentary manner. In this ICCV 2015 paper, we present a novel pipeline built upon deep CNN features to harvest discriminative visual objects and parts for scene classification. We first use a region proposal technique to generate a set of high-quality patches potentially containing objects, and apply a pre-trained CNN to extract generic deep features from these patches. Then we perform both unsupervised and weakly supervised learning to screen these patches and discover discriminative ones representing category-specific objects and parts. We further apply discriminative clustering enhanced with local CNN fine-tuning to aggregate similar objects and parts into groups, called meta objects. A scene image representation is constructed by pooling the feature response maps of all the learned meta objects at multiple spatial scales. We have confirmed that the scene image representation obtained using this new pipeline is capable of delivering state-of-the-art performance on two popular scene benchmark datasets, MIT Indoor 67~\cite{MITIndoor67} and Sun397~\cite{Sun397}
Accurate representational learning of both the explicit and implicit relationships within data is critical to the ability of machines to perform more complex and abstract reasoning tasks. We describe the efficient weakly supervised learning of such inferences by our Dynamic Adaptive Network Intelligence (DANI) model. We report state-of-the-art results for DANI over question answering tasks in the bAbI dataset that have proved difficult for contemporary approaches to learning representation (Weston et al., 2015).
Recent advances of deep learning have achieved remarkable performances in various challenging computer vision tasks. Especially in object localization, deep convolutional neural networks outperform traditional approaches based on extraction of data/task-driven features instead of hand-crafted features. Although location information of region-of-interests (ROIs) gives good prior for object localization, it requires heavy annotation efforts from human resources. Thus a weakly supervised framework for object localization is introduced. The term "weakly" means that this framework only uses image-level labeled datasets to train a network. With the help of transfer learning which adopts weight parameters of a pre-trained network, the weakly supervised learning framework for object localization performs well because the pre-trained network already has well-trained class-specific features. However, those approaches cannot be used for some applications which do not have pre-trained networks or well-localized large scale images. Medical image analysis is a representative among those applications because it is impossible to obtain such pre-trained networks. In this work, we present a "fully" weakly supervised framework for object localization ("semi"-weakly is the counterpart which uses pre-trained filters for weakly supervised localization) named as self-transfer learning (STL). It jointly optimizes both classification and localization networks simultaneously. By controlling a supervision level of the localization network, STL helps the localization network focus on correct ROIs without any types of priors. We evaluate the proposed STL framework using two medical image datasets, chest X-rays and mammograms, and achieve signiticantly better localization performance compared to previous weakly supervised approaches.
We prove that the empirical risk of most well-known loss functions factors into a linear term aggregating all labels with a term that is label free, and can further be expressed by sums of the loss. This holds true even for non-smooth, non-convex losses and in any RKHS. The first term is a (kernel) mean operator --the focal quantity of this work-- which we characterize as the sufficient statistic for the labels. The result tightens known generalization bounds and sheds new light on their interpretation. Factorization has a direct application on weakly supervised learning. In particular, we demonstrate that algorithms like SGD and proximal methods can be adapted with minimal effort to handle weak supervision, once the mean operator has been estimated. We apply this idea to learning with asymmetric noisy labels, connecting and extending prior work. Furthermore, we show that most losses enjoy a data-dependent (by the mean operator) form of noise robustness, in contrast with known negative results.