In this study we investigate the potential for using synthetic aperture radar (SAR) data to provide high resolution defoliation and regrowth mapping of trees in the tundra-forest ecotone. Using aerial photographs, four areas with live forest and four areas with dead trees were identified. Quad-polarimetric SAR data from RADARSAT-2 was collected from the same area, and the complex multilook polarimetric covariance matrix was calculated using a novel extension of guided nonlocal means speckle filtering. The nonlocal approach allows us to preserve the high spatial resolution of single-look complex data, which is essential for accurate mapping of the sparsely scattered trees in the study area. Using a standard random forest classification algorithm, our filtering results in over $99.7 \%$ classification accuracy, higher than traditional speckle filtering methods, and on par with the classification accuracy based on optical data.
We develop a novel method, called PoWER-BERT, for improving the inference time of the popular BERT model, while maintaining the accuracy. It works by: a) exploiting redundancy pertaining to word-vectors (intermediate encoder outputs) and eliminating the redundant vectors. b) determining which word-vectors to eliminate by developing a strategy for measuring their significance, based on the self-attention mechanism. c) learning how many word-vectors to eliminate by augmenting the BERT model and the loss function. Experiments on the standard GLUE benchmark shows that PoWER-BERT achieves up to 4.5x reduction in inference time over BERT with <1% loss in accuracy. We show that PoWER-BERT offers significantly better trade-off between accuracy and inference time compared to prior methods. We demonstrate that our method attains up to 6.8x reduction in inference time with <1% loss in accuracy when applied over ALBERT, a highly compressed version of BERT. The code for PoWER-BERT is publicly available at https://github.com/IBM/PoWER-BERT.
Conversational agents such as Cortana, Alexa and Siri are continuously working on increasing their capabilities by adding new domains. The support of a new domain includes the design and development of a number of NLU components for domain classification, intents classification and slots tagging (including named entity recognition). Each component only performs well when trained on a large amount of labeled data. Second, these components are deployed on limited-memory devices which requires some model compression. Third, for some domains such as the health domain, it is hard to find a single training data set that covers all the required slot types. To overcome these mentioned problems, we present a multi-task transformer-based neural architecture for slot tagging. We consider the training of a slot tagger using multiple data sets covering different slot types as a multi-task learning problem. The experimental results on the biomedical domain have shown that the proposed approach outperforms the previous state-of-the-art systems for slot tagging on the different benchmark biomedical datasets in terms of (time and memory) efficiency and effectiveness. The output slot tagger can be used by the conversational agent to better identify entities in the input utterances.
Anomaly detection algorithms find extensive use in various fields. This area of research has recently made great advances thanks to deep learning. A recent method, the deep Support Vector Data Description (deep SVDD), which is inspired by the classic kernel-based Support Vector Data Description (SVDD), is capable of simultaneously learning a feature representation of the data and a data-enclosing hypersphere. The method has shown promising results in both unsupervised and semi-supervised settings. However, deep SVDD suffers from hypersphere collapse -- also known as mode collapse, if the architecture of the model does not comply with certain architectural constraints, e.g. the removal of bias terms. These constraints limit the adaptability of the model and in some cases, may affect the model performance due to learning sub-optimal features. In this work, we consider two regularizers to prevent hypersphere collapse in deep SVDD. The first regularizer is based on injecting random noise via the standard cross-entropy loss. The second regularizer penalizes the minibatch variance when it becomes too small. Moreover, we introduce an adaptive weighting scheme to control the amount of penalization between the SVDD loss and the respective regularizer. Our proposed regularized variants of deep SVDD show encouraging results and outperform a prominent state-of-the-art method on a setup where the anomalies have no apparent geometrical structure.
The automatic detection of critical findings in chest X-rays (CXR), such as pneumothorax, is important for assisting radiologists in their clinical workflow like triaging time-sensitive cases and screening for incidental findings. While deep learning (DL) models has become a promising predictive technology with near-human accuracy, they commonly suffer from a lack of explainability, which is an important aspect for clinical deployment of DL models in the highly regulated healthcare industry. For example, localizing critical findings in an image is useful for explaining the predictions of DL classification algorithms. While there have been a host of joint classification and localization methods for computer vision, the state-of-the-art DL models require locally annotated training data in the form of pixel level labels or bounding box coordinates. In the medical domain, this requires an expensive amount of manual annotation by medical experts for each critical finding. This requirement becomes a major barrier for training models that can rapidly scale to various findings. In this work, we address these shortcomings with an interpretable DL algorithm based on multi-instance learning that jointly classifies and localizes critical findings in CXR without the need for local annotations. We show competitive classification results on three different critical findings (pneumothorax, pneumonia, and pulmonary edema) from three different CXR datasets.
The improvements in magnetic resonance imaging have led to the development of numerous techniques to better detect structural alterations caused by neurodegenerative diseases. Among these, the patch-based grading framework has been proposed to model local patterns of anatomical changes. This approach is attractive because of its low computational cost and its competitive performance. Other studies have proposed to analyze the deformations of brain structures using tensor-based morphometry, which is a highly interpretable approach. In this work, we propose to combine the advantages of these two approaches by extending the patch-based grading framework with a new tensor-based grading method that enables us to model patterns of local deformation using a log-Euclidean metric. We evaluate our new method in a study of the putamen for the classification of patients with pre-manifest Huntington's disease and healthy controls. Our experiments show a substantial increase in classification accuracy (87.5 $\pm$ 0.5 vs. 81.3 $\pm$ 0.6) compared to the existing patch-based grading methods, and a good complement to putamen volume, which is a primary imaging-based marker for the study of Huntington's disease.
To decrease patient waiting time for diagnosis of the Coronary Artery Disease, automatic methods are applied to identify its severity using Coronary Computed Tomography Angiography scans or extracted Multiplanar Reconstruction (MPR) images, giving doctors a second-opinion on the priority of each case. The main disadvantage of previous studies is the lack of large set of data that could guarantee their reliability. Another limitation is the usage of handcrafted features requiring manual preprocessing, such as centerline extraction. We overcome both limitations by applying a different automated approach based on ShuffleNet V2 network architecture and testing it on the proposed collected dataset of MPR images, which is bigger than any other used in this field before. We also omit centerline extraction step and train and test our model using whole curved MPR images of 708 and 105 patients, respectively. The model predicts one of three classes: 'no stenosis' for normal, 'non-significant' - 1-50% of stenosis detected, 'significant' - more than 50% of stenosis. We demonstrate model's interpretability through visualization of the most important features selected by the network. For stenosis score classification, the method shows improved performance comparing to previous works, achieving 80% accuracy on the patient level. Our code is publicly available.
Few-shot classification aims to recognize novel categories with only few labeled images in each class. Existing metric-based few-shot classification algorithms predict categories by comparing the feature embeddings of query images with those from a few labeled images (support examples) using a learned metric function. While promising performance has been demonstrated, these methods often fail to generalize to unseen domains due to large discrepancy of the feature distribution across domains. In this work, we address the problem of few-shot classification under domain shifts for metric-based methods. Our core idea is to use feature-wise transformation layers for augmenting the image features using affine transforms to simulate various feature distributions under different domains in the training stage. To capture variations of the feature distributions under different domains, we further apply a learning-to-learn approach to search for the hyper-parameters of the feature-wise transformation layers. We conduct extensive experiments and ablation studies under the domain generalization setting using five few-shot classification datasets: mini-ImageNet, CUB, Cars, Places, and Plantae. Experimental results demonstrate that the proposed feature-wise transformation layer is applicable to various metric-based models, and provides consistent improvements on the few-shot classification performance under domain shift.
Deep neural networks (DNNs) have recently received vast attention in applications requiring classification of radar returns, including radar-based human activity recognition for security, smart homes, assisted living, and biomedicine. However,acquiring a sufficiently large training dataset remains a daunting task due to the high human costs and resources required for radar data collection. In this paper, an extended approach to adversarial learning is proposed for generation of synthetic radar micro-Doppler signatures that are well-adapted to different environments. The synthetic data is evaluated using visual interpretation, analysis of kinematic consistency, data diversity, dimensions of the latent space, and saliency maps. A principle-component analysis (PCA) based kinematic-sifting algorithm is introduced to ensure that synthetic signatures are consistent with physically possible human motions. The synthetic dataset is used to train a 19-layer deep convolutional neural network (DCNN) to classify micro-Doppler signatures acquired from an environment different from that of the dataset supplied to the adversarial network. An overall accuracy 93% is achieved on a dataset that contains multiple aspect angles (0 deg., 30 deg., and 45 deg. as well as 60 deg.), with 9% improvement as a result of kinematic sifting.
The goal of few-shot learning is to learn a model that can recognize novel classes based on one or few training data. It is challenging mainly due to two aspects: (1) it lacks good feature representation of novel classes; (2) a few of labeled data could not accurately represent the true data distribution and thus it's hard to learn a good decision function for classification. In this work, we use a sophisticated network architecture to learn better feature representation and focus on the second issue. A novel continual local replacement strategy is proposed to address the data deficiency problem. It takes advantage of the content in unlabeled images to continually enhance labeled ones. Specifically, a pseudo labeling method is adopted to constantly select semantically similar images on the fly. Original labeled images will be locally replaced by the selected images for the next epoch training. In this way, the model can directly learn new semantic information from unlabeled images and the capacity of supervised signals in the embedding space can be significantly enlarged. This allows the model to improve generalization and learn a better decision boundary for classification. Our method is conceptually simple and easy to implement. Extensive experiments demonstrate that it can achieve state-of-the-art results on various few-shot image recognition benchmarks.
We present an algorithm for supervised learning using tensor networks, employing a step of preprocessing the data by coarse-graining through a sequence of wavelet transformations. We represent these transformations as a set of tensor network layers identical to those in a multi-scale entanglement renormalization ansatz (MERA) tensor network, and perform supervised learning and regression tasks through a model based on a matrix product state (MPS) tensor network acting on the coarse-grained data. Because the entire model consists of tensor contractions (apart from the initial non-linear feature map), we can adaptively fine-grain the optimized MPS model backwards through the layers with essentially no loss in performance. The MPS itself is trained using an adaptive algorithm based on the density matrix renormalization group (DMRG) algorithm. We test our methods by performing a classification task on audio data and a regression task on temperature time-series data, studying the dependence of training accuracy on the number of coarse-graining layers and showing how fine-graining through the network may be used to initialize models with access to finer-scale features.
We propose a representation learning framework for medical diagnosis domain. It is based on heterogeneous network-based model of diagnostic data as well as modified metapath2vec algorithm for learning latent node representation. We compare the proposed algorithm with other representation learning methods in two practical case studies: symptom/disease classification and disease prediction. We observe a significant performance boost in these task resulting from learning representations of domain data in a form of heterogeneous network.
Uncertainty quantification for deep learning is a challenging open problem. Bayesian statistics offer a mathematically grounded framework to reason about uncertainties; however, approximate posteriors for modern neural networks still require prohibitive computational costs. We propose a family of algorithms which split the classification task into two stages: representation learning and uncertainty estimation. We compare four specific instances, where uncertainty estimation is performed via either an ensemble of Stochastic Gradient Descent or Stochastic Gradient Langevin Dynamics snapshots, an ensemble of bootstrapped logistic regressions, or via a number of Monte Carlo Dropout passes. We evaluate their performance in terms of \emph{selective} classification (risk-coverage), and their ability to detect out-of-distribution samples. Our experiments suggest there is limited value in adding multiple uncertainty layers to deep classifiers, and we observe that these simple methods strongly outperform a vanilla point-estimate SGD in some complex benchmarks like ImageNet.
A person ontology comprising concepts, attributes and relationships of people has a number of applications in data protection, didentification, population of knowledge graphs for business intelligence and fraud prevention. While artificial neural networks have led to improvements in Entity Recognition, Entity Classification, and Relation Extraction, creating an ontology largely remains a manual process, because it requires a fixed set of semantic relations between concepts. In this work, we present a system for automatically populating a person ontology graph from unstructured data using neural models for Entity Classification and Relation Extraction. We introduce a new dataset for these tasks and discuss our results.
To learn models or features that generalize across tasks and domains is one of the grand goals of machine learning. In this paper, we propose to use cross-domain, cross-task data as validation objective for hyper-parameter optimization (HPO) to improve on this goal. Given a rich enough search space, optimization of hyper-parameters learn features that maximize validation performance and, due to the objective, generalize across tasks and domains. We demonstrate the effectiveness of this strategy on few-shot image classification within and across domains. The learned features outperform all previous few-shot and meta-learning approaches.
Existing graph neural networks may suffer from the "suspended animation problem" when the model architecture goes deep. Meanwhile, for some graph learning scenarios, e.g., nodes with text/image attributes or graphs with long-distance node correlations, deep graph neural networks will be necessary for effective graph representation learning. In this paper, we propose a new graph neural network, namely DIFNET (Graph Diffusive Neural Network), for graph representation learning and node classification. DIFNET utilizes both neural gates and graph residual learning for node hidden state modeling, and includes an attention mechanism for node neighborhood information diffusion. Extensive experiments will be done in this paper to compare DIFNET against several state-of-the-art graph neural network models. The experimental results can illustrate both the learning performance advantages and effectiveness of DIFNET, especially in addressing the "suspended animation problem".
This paper proposes a novel approach for unsupervised domain adaptation (UDA) with target shift. Target shift is a problem of mismatch in label distribution between source and target domains. Typically it appears as class-imbalance in target domain. In practice, this is an important problem in UDA; as we do not know labels in target domain datasets, we do not know whether or not its distribution is identical to that in the source domain dataset. Many traditional approaches achieve UDA with distribution matching by minimizing mean maximum discrepancy or adversarial training; however these approaches implicitly assume a coincidence in the distributions and do not work under situations with target shift. Some recent UDA approaches focus on class boundary and some of them are robust to target shift, but they are only applicable to classification and not to regression. To overcome the target shift problem in UDA, the proposed method, partially shared variational autoencoders (PS-VAEs), uses pair-wise feature alignment instead of feature distribution matching. PS-VAEs inter-convert domain of each sample by a CycleGAN-based architecture while preserving its label-related content. To evaluate the performance of PS-VAEs, we carried out two experiments: UDA with class-unbalanced digits datasets (classification), and UDA from synthesized data to real observation in human-pose-estimation (regression). The proposed method presented its robustness against the class-imbalance in the classification task, and outperformed the other methods in the regression task with a large margin.
A tensor provides a concise way to codify the interdependence of complex data. Treating a tensor as a d-way array, each entry records the interaction between the different indices. Clustering provides a way to parse the complexity of the data into more readily understandable information. Clustering methods are heavily dependent on the algorithm of choice, as well as the chosen hyperparameters of the algorithm. However, their sensitivity to data scales is largely unknown. In this work, we apply the discrete wavelet transform to analyze the effects of coarse-graining on clustering tensor data. We are particularly interested in understanding how scale effects clustering of the Earth's climate system. The discrete wavelet transform allows classification of the Earth's climate across a multitude of spatial-temporal scales. The discrete wavelet transform is used to produce an ensemble of classification estimates, as opposed to a single classification. Information theoretic approaches are used to identify important scale lenghts in clustering The L15 Climate Datset. We also discover a sub-collection of the ensemble that spans the majority of the variance observed, allowing for efficient consensus clustering techniques that can be used to identify climate biomes.
In vision-based object classification systems imaging sensors perceive the environment and machine learning is then used to detect and classify objects for decision-making purposes; e.g., to maneuver an automated vehicle around an obstacle or to raise an alarm to indicate the presence of an intruder in surveillance settings. In this work we demonstrate how the perception domain can be remotely and unobtrusively exploited to enable an attacker to create spurious objects or alter an existing object. An automated system relying on a detection/classification framework subject to our attack could be made to undertake actions with catastrophic results due to attacker-induced misperception. We focus on camera-based systems and show that it is possible to remotely project adversarial patterns into camera systems by exploiting two common effects in optical imaging systems, viz., lens flare/ghost effects and auto-exposure control. To improve the robustness of the attack to channel effects, we generate optimal patterns by integrating adversarial machine learning techniques with a trained end-to-end channel model. We experimentally demonstrate our attacks using a low-cost projector, on three different image datasets, in indoor and outdoor environments, and with three different cameras. Experimental results show that, depending on the projector-camera distance, attack success rates can reach as high as 100% and under targeted conditions.
Due to the monumental growth of Internet applications in the last decade, the need for security of information network has increased manifolds. As a primary defense of network infrastructure, an intrusion detection system is expected to adapt to dynamically changing threat landscape. Many supervised and unsupervised techniques have been devised by researchers from the discipline of machine learning and data mining to achieve reliable detection of anomalies. Deep learning is an area of machine learning which applies neuron-like structure for learning tasks. Deep learning has profoundly changed the way we approach learning tasks by delivering monumental progress in different disciplines like speech processing, computer vision, and natural language processing to name a few. It is only relevant that this new technology must be investigated for information security applications. The aim of this paper is to investigate the suitability of deep learning approaches for anomaly-based intrusion detection system. For this research, we developed anomaly detection models based on different deep neural network structures, including convolutional neural networks, autoencoders, and recurrent neural networks. These deep models were trained on NSLKDD training data set and evaluated on both test data sets provided by NSLKDD, namely NSLKDDTest+ and NSLKDDTest21. All experiments in this paper are performed by authors on a GPU-based test bed. Conventional machine learning-based intrusion detection models were implemented using well-known classification techniques, including extreme learning machine, nearest neighbor, decision-tree, random-forest, support vector machine, naive-bays, and quadratic discriminant analysis. Both deep and conventional machine learning models were evaluated using well-known classification metrics, including receiver operating characteristics, area under curve, precision-recall curve, mean average precision and accuracy of classification. Experimental results of deep IDS models showed promising results for real-world application in anomaly detection systems.