While training on samples drawn from independent and identical distribution has been a de facto paradigm for optimizing image classification networks, humans learn new concepts in an easy-to-hard manner and on the selected examples progressively. Driven by this fact, we investigate the training paradigms where the samples are not drawn from independent and identical distribution. We propose a data sampling strategy, named Drop-and-Refresh (DaR), motivated by the learning behaviors of humans that selectively drop easy samples and refresh them only periodically. We show in our experiments that the proposed DaR strategy can maintain (and in many cases improve) the predictive accuracy even when the training cost is reduced by 15% on various datasets (CIFAR 10, CIFAR 100 and ImageNet) and with different backbone architectures (ResNets, DenseNets and MobileNets). Furthermore and perhaps more importantly, we find the ImageNet pre-trained models using our DaR sampling strategy achieves better transferability for the downstream tasks including object detection (+0.3 AP), instance segmentation (+0.3 AP), scene parsing (+0.5 mIoU) and human pose estimation (+0.6 AP). Our investigation encourages people to rethink the connections between the sampling strategy for training and the transferability of its learned features for pre-training ImageNet models.
Understanding driver activity is vital for in-vehicle systems that aim to reduce the incidence of car accidents rooted in cognitive distraction. Automating real-time behavior recognition while ensuring actions classification with high accuracy is however challenging, given the multitude of circumstances surrounding drivers, the unique traits of individuals, and the computational constraints imposed by in-vehicle embedded platforms. Prior work fails to jointly meet these runtime/accuracy requirements and mostly rely on a single sensing modality, which in turn can be a single point of failure. In this paper, we harness the exceptional feature extraction abilities of deep learning and propose a dedicated Interwoven Deep Convolutional Neural Network (InterCNN) architecture to tackle the problem of accurate classification of driver behaviors in real-time. The proposed solution exploits information from multi-stream inputs, i.e., in-vehicle cameras with different fields of view and optical flows computed based on recorded images, and merges through multiple fusion layers abstract features that it extracts. This builds a tight ensembling system, which significantly improves the robustness of the model. In addition, we introduce a temporal voting scheme based on historical inference instances, to enhance the classification accuracy. Experiments conducted with a dataset that we collect in a mock-up car environment demonstrate that the proposed InterCNN with MobileNet convolutional blocks can classify 9 different behaviors with 73.97% accuracy, and 5 'aggregated' behaviors with 81.66% accuracy. We further show that our architecture is highly computationally efficient, as it performs inferences within 15ms, which satisfies the real-time constraints of intelligent cars. Nevertheless, our InterCNN is robust to lossy input, as the classification remains accurate when two input streams are occluded.
Neural network based classifiers are still prone to manipulation through adversarial perturbations. State of the art attacks can overcome most of the defense or detection mechanisms suggested so far, and adversaries have the upper hand in this arms race. Adversarial examples are designed to resemble the normal input from which they were constructed, while triggering an incorrect classification. This basic design goal leads to a characteristic spatial behavior within the context of Activation Spaces, a term coined by the authors to refer to the hyperspaces formed by the activation values of the network's layers. Within the output of the first layers of the network, an adversarial example is likely to resemble normal instances of the source class, while in the final layers such examples will diverge towards the adversary's target class. The steps below enable us to leverage this inherent shift from one class to another in order to form a novel adversarial example detector. We construct Euclidian spaces out of the activation values of each of the deep neural network layers. Then, we induce a set of k-nearest neighbor classifiers (k-NN), one per activation space of each neural network layer, using the non-adversarial examples. We leverage those classifiers to produce a sequence of class labels for each nonperturbed input sample and estimate the a priori probability for a class label change between one activation space and another. During the detection phase we compute a sequence of classification labels for each input using the trained classifiers. We then estimate the likelihood of those classification sequences and show that adversarial sequences are far less likely than normal ones. We evaluated our detection method against the state of the art C&W attack method, using two image classification datasets (MNIST, CIFAR-10) reaching an AUC 0f 0.95 for the CIFAR-10 dataset.
In this work, we describe our approach to pneumonia classification and localization in chest radiographs. This method uses only \emph{open-source} deep learning object detection and is based on CoupleNet, a fully convolutional network which incorporates global and local features for object detection. Our approach achieves robustness through critical modifications of the training process and a novel ensembling algorithm which merges bounding boxes from several models. We tested our detection algorithm tested on a dataset of 3000 chest radiographs as part of the 2018 RSNA Pneumonia Challenge; our solution was recognized as a winning entry in a contest which attracted more than 1400 participants worldwide.
The classification of electrocardiographic (ECG) signals is a challenging problem for healthcare industry. Traditional supervised learning methods require a large number of labeled data which is usually expensive and difficult to obtain for ECG signals. Active learning is well-suited for ECG signal classification as it aims at selecting the best set of labeled data in order to maximize the classification performance. Motivated by the fact that ECG data are usually heavily unbalanced among different classes and the class labels are noisy as they are manually labeled, this paper proposes a novel solution based on robust active learning for addressing these challenges. The key idea is to first apply the clustering of the data in a low dimensional embedded space and then select the most information instances within local clusters. By selecting the most informative instances relying on local average minimal distances, the algorithm tends to select the data for labelling in a more diversified way. Finally, the robustness of the model is further enhanced by adding a novel noisy label reduction scheme after the selection of the labeled data. Experiments on the ECG signal classification from the MIT-BIH arrhythmia database demonstrate the effectiveness of the proposed algorithm.
Spectral Graph Convolutional Networks (GCNs) are a generalization of convolutional networks to learning on graph-structured data. Applications of spectral GCNs have been successful, but limited to a few problems where the graph is fixed, such as shape correspondence and node classification. In this work, we address this limitation by revisiting a particular family of spectral graph networks, Chebyshev GCNs, showing its efficacy in solving graph classification tasks with a variable graph structure and size. Chebyshev GCNs restrict graphs to have at most one edge between any pair of nodes. To this end, we propose a novel multigraph network that learns from multi-relational graphs. We model learned edges with abstract meaning and experiment with different ways to fuse the representations extracted from annotated and learned edges, achieving competitive results on a variety of chemical classification benchmarks.
We consider a high dimensional binary classification problem and construct a classification procedure by minimizing the empirical misclassification risk with a penalty on the number of selected features. We derive non-asymptotic probability bounds on the estimated sparsity as well as on the excess misclassification risk. In particular, we show that our method yields a sparse solution whose l0-norm can be arbitrarily close to true sparsity with high probability and obtain the rates of convergence for the excess misclassification risk. The proposed procedure is implemented via the method of mixed integer linear programming. Its numerical performance is illustrated in Monte Carlo experiments.
In the biomedical domain, the lack of sharable datasets often limit the possibility of developing natural language processing systems, especially dialogue applications and natural language understanding models. To overcome this issue, we explore data generation using templates and terminologies and data augmentation approaches. Namely, we report our experiments using paraphrasing and word representations learned on a large EHR corpus with Fasttext and ELMo, to learn a NLU model without any available dataset. We evaluate on a NLU task of natural language queries in EHRs divided in slot-filling and intent classification sub-tasks. On the slot-filling task, we obtain a F-score of 0.76 with the ELMo representation; and on the classification task, a mean F-score of 0.71. Our results show that this method could be used to develop a baseline system.
Text classification is one of the fundamental tasks in natural language processing. Recently, deep neural networks have achieved promising performance in the text classification task compared to shallow models. Despite of the significance of deep models, they ignore the fine-grained (matching signals between words and classes) classification clues since their classifications mainly rely on the text-level representations. To address this problem, we introduce the interaction mechanism to incorporate word-level matching signals into the text classification task. In particular, we design a novel framework, EXplicit interAction Model (dubbed as EXAM), equipped with the interaction mechanism. We justified the proposed approach on several benchmark datasets including both multi-label and multi-class text classification tasks. Extensive experimental results demonstrate the superiority of the proposed method. As a byproduct, we have released the codes and parameter settings to facilitate other researches.
Entity Type Classification can be defined as the task of assigning category labels to entity mentions in documents. While neural networks have recently improved the classification of general entity mentions, pattern matching and other systems continue to be used for classifying personal data entities (e.g. classifying an organization as a media company or a government institution for GDPR, and HIPAA compliance). We propose a neural model to expand the class of personal data entities that can be classified at a fine grained level, using the output of existing pattern matching systems as additional contextual features. We introduce new resources, a personal data entities hierarchy with 134 types, and two datasets from the Wikipedia pages of elected representatives and Enron emails. We hope these resource will aid research in the area of personal data discovery, and to that effect, we provide baseline results on these datasets, and compare our method with state of the art models on OntoNotes dataset.
Making a simple model by choosing a limited number of features with the purpose of reducing the computational complexity of the algorithms involved in classification is one of the main issues in machine learning and data mining. The aim of Feature Selection (FS) is to reduce the number of redundant and irrelevant features and improve the accuracy of classification in a data set. We propose an efficient ISPSO-GLOBAL (Improved Seeding Particle Swarm Optimization GLOBAL) method which investigates the specified iterations to produce prominent features and store them in storage list. The goal is to find informative features based on its iteration frequency with favorable fitness for the next generation and high exploration. Our method exploits of a new initialization strategy in PSO which improves space search and utilizes chaos theory to enhance the population initialization, then we offer a new formula to determine the features size used in proposed method. Our experiments with real-world data sets show that the performance of the ISPSO-GLOBAL is superior comparing with state-of-the-art methods in most of the data sets.
Integrated gradients are widely employed to evaluate the contribution of input features in classification models because it satisfies the axioms for attribution of prediction. This method, however, requires an appropriate baseline for reliable determination of the contributions. We propose a compensated integrated gradients method that does not require a baseline. In fact, the method compensates the attributions calculated by integrated gradients at an arbitrary baseline using Shapley sampling. We prove that the method retrieves reliable attributions if the processes of input features in a classifier are mutually independent, and they are identical like shared weights in convolutional neural networks. Using three electroencephalogram datasets, we experimentally demonstrate that the attributions of the proposed method are more reliable than those of the original integrated gradients, and its computational complexity is much lower than that of Shapley sampling.
Recently, a large number of neural mechanisms and models have been proposed for sequence learning, of which self-attention, as exemplified by the Transformer model, and graph neural networks (GNNs) have attracted much attention. In this paper, we propose an approach that combines and draws on the complementary strengths of these two methods. Specifically, we propose contextualized non-local neural networks (CN$^{\textbf{3}}$), which can both dynamically construct a task-specific structure of a sentence and leverage rich local dependencies within a particular neighborhood. Experimental results on ten NLP tasks in text classification, semantic matching, and sequence labeling show that our proposed model outperforms competitive baselines and discovers task-specific dependency structures, thus providing better interpretability to users.
In real-world recognition/classification tasks, limited by various objective factors, it is usually difficult to collect training samples to exhaust all classes when training a recognizer or classifier. A more realistic scenario is open set recognition (OSR), where incomplete knowledge of the world exists at training time, and unknown classes can be submitted to an algorithm during testing, requiring the classifiers to not only accurately classify the seen classes, but also effectively deal with the unseen ones. This paper provides a comprehensive survey of existing open set recognition techniques covering various aspects ranging from related definitions, representations of models, datasets, evaluation criteria, and algorithm comparisons. Furthermore, we briefly analyze the relationships between OSR and its related tasks including zero-shot, one-shot (few-shot) recognition/learning techniques, classification with reject option, and so forth. Additionally, we also overview the open world recognition which can be seen as a natural extension of OSR. Importantly, we highlight the limitations of existing approaches and point out some promising subsequent research directions in this field.
Deep learning-based methods, such as the sliding window approach for cropped-image classification and heuristic aggregation for whole-slide inference, for analyzing histological patterns in high-resolution microscopy images have shown promising results. These approaches, however, require a laborious annotation process and are fragmented. This diagnostic study collected deidentified high-resolution histological images (N = 379) for training a new model composed of a convolutional neural network and a grid-based attention network, trainable without region-of-interest annotations. Histological images of patients who underwent endoscopic esophagus and gastroesophageal junction mucosal biopsy between January 1, 2016, and December 31, 2018, at Dartmouth-Hitchcock Medical Center (Lebanon, New Hampshire) were collected. The method achieved a mean accuracy of 0.83 in classifying 123 test images. These results were comparable with or better than the performance from the current state-of-the-art sliding window approach, which was trained with regions of interest. Results of this study suggest that the proposed attention-based deep neural network framework for Barrett esophagus and esophageal adenocarcinoma detection is important because it is based solely on tissue-level annotations, unlike existing methods that are based on regions of interest. This new model is expected to open avenues for applying deep learning to digital pathology.
Multi-view data, that is matched sets of measurements on the same subjects, have become increasingly common with advances in multi-omics technology. Often, it is of interest to find associations between the views that are related to the intrinsic class memberships. Existing association methods cannot directly incorporate class information, while existing classification methods do not take into account between-views associations. In this work, we propose a framework for Joint Association and Classification Analysis of multi-view data (JACA). Our goal is not to merely improve the misclassification rates, but to provide a latent representation of high-dimensional data that is both relevant for the subtype discrimination and coherent across the views. We motivate the methodology by establishing a connection between canonical correlation analysis and discriminant analysis. We also establish the estimation consistency of JACA in high-dimensional settings. A distinct advantage of JACA is that it can be applied to the multi-view data with block-missing structure, that is to cases where a subset of views or class labels is missing for some subjects. The application of JACA to quantify the associations between RNAseq and miRNA views with respect to consensus molecular subtypes in colorectal cancer data from The Cancer Genome Atlas project leads to improved misclassification rates and stronger found associations compared to existing methods.
X-ray diffraction (XRD) data acquisition and analysis is among the most time-consuming steps in the development cycle of novel thin-film materials. We propose a machine-learning-enabled approach to predict crystallographic dimensionality and space group from a limited number of thin-film XRD patterns. We overcome the scarce-data problem intrinsic to novel materials development by coupling a supervised machine learning approach with a model agnostic, physics-informed data augmentation strategy using simulated data from the Inorganic Crystal Structure Database (ICSD) and experimental data. As a test case, 115 thin-film metal halides spanning 3 dimensionalities and 7 space-groups are synthesized and classified. After testing various algorithms, we develop and implement an all convolutional neural network, with cross validated accuracies for dimensionality and space-group classification of 93% and 89%, respectively. We propose average class activation maps, computed from a global average pooling layer, to allow high model interpretability by human experimentalists, elucidating the root causes of misclassification. Finally, we systematically evaluate the maximum XRD pattern step size (data acquisition rate) before loss of predictive accuracy occurs, and determine it to be 0.16{\deg}, which enables an XRD pattern to be obtained and classified in 5.5 minutes or less.
Adversarial approach has been widely used for data generation in the last few years. However, this approach has not been extensively utilized for classifier training. In this paper, we propose an adversarial framework for classifier training that can also handle imbalanced data. Indeed, a network is trained via an adversarial approach to give weights to samples of the majority class such that the obtained classification problem becomes more challenging for the discriminator and thus boosts its classification capability. In addition to the general imbalanced classification problems, the proposed method can also be used for problems such as graph representation learning in which it is desired to discriminate similar nodes from dissimilar nodes. Experimental results on imbalanced data classification and on the tasks like graph link prediction show the superiority of the proposed method compared to the state-of-the-art methods.
Graph embedding is an important approach for graph analysis tasks such as node classification and link prediction. The goal of graph embedding is to find a low dimensional representation of graph nodes that preserves the graph information. Recent methods like Graph Convolutional Network (GCN) try to consider node attributes (if available) besides node relations and learn node embeddings for unsupervised and semi-supervised tasks on graphs. On the other hand, multi-layer graph analysis has been received attention recently. However, the existing methods for multi-layer graph embedding cannot incorporate all available information (like node attributes). Moreover, most of them consider either type of nodes or type of edges, and they do not treat within and between layer edges differently. In this paper, we propose a method called MGCN that utilizes the GCN for multi-layer graphs. MGCN embeds nodes of multi-layer graphs using both within and between layers relations and nodes attributes. We evaluate our method on the semi-supervised node classification task. Experimental results demonstrate the superiority of the proposed method to other multi-layer and single-layer competitors and also show the positive effect of using cross-layer edges.
Ranking functions in information retrieval are often used in search engines to recommend the relevant answers to the query. This paper makes use of this notion of information retrieval and applies onto the problem domain of cognate detection. The main contributions of this paper are: (1) positional segmentation, which incorporates the sequential notion; (2) graphical error modelling, which deduces the transformations. The current research work focuses on classification problem; which is distinguishing whether a pair of words are cognates. This paper focuses on a harder problem, whether we could predict a possible cognate from the given input. Our study shows that when language modelling smoothing methods are applied as the retrieval functions and used in conjunction with positional segmentation and error modelling gives better results than competing baselines, in both classification and prediction of cognates. Source code is at: https://github.com/pranav-ust/cognates