Learning algorithms that aggregate predictions from an ensemble of diverse base classifiers consistently outperform individual methods. Many of these strategies have been developed in a supervised setting, where the accuracy of each base classifier can be empirically measured and this information is incorporated in the training process. However, the reliance on labeled data precludes the application of ensemble methods to many real world problems where labeled data has not been curated. To this end we developed a new theoretical framework for binary classification, the Strategy for Unsupervised Multiple Method Aggregation (SUMMA), to estimate the performances of base classifiers and an optimal strategy for ensemble learning from unlabeled data.
In this work we tackle the problem of sentence boundary detection applied to French as a binary classification task ("sentence boundary" or "not sentence boundary"). We combine convolutional neural networks with subword-level information vectors, which are word embedding representations learned from Wikipedia that take advantage of the words morphology; so each word is represented as a bag of their character n-grams. We decide to use a big written dataset (French Gigaword) instead of standard size transcriptions to train and evaluate the proposed architectures with the intention of using the trained models in posterior real life ASR transcriptions. Three different architectures are tested showing similar results; general accuracy for all models overpasses 0.96. All three models have good F1 scores reaching values over 0.97 regarding the "not sentence boundary" class. However, the "sentence boundary" class reflects lower scores decreasing the F1 metric to 0.778 for one of the models. Using subword-level information vectors seem to be very effective leading to conclude that the morphology of words encoded in the embeddings representations behave like pixels in an image making feasible the use of convolutional neural network architectures.
As a basic task in computer vision, semantic segmentation can provide fundamental information for object detection and instance segmentation to help the artificial intelligence better understand real world. Since the proposal of fully convolutional neural network (FCNN), it has been widely used in semantic segmentation because of its high accuracy of pixel-wise classification as well as high precision of localization. In this paper, we apply several famous FCNN to brain tumor segmentation, making comparisons and adjusting network architectures to achieve better performance measured by metrics such as precision, recall, mean of intersection of union (mIoU) and dice score coefficient (DSC). The adjustments to the classic FCNN include adding more connections between convolutional layers, enlarging decoders after up sample layers and changing the way shallower layers' information is reused. Besides the structure modification, we also propose a new classifier with a hierarchical dice loss. Inspired by the containing relationship between classes, the loss function converts multiple classification to multiple binary classification in order to counteract the negative effect caused by imbalance data set. Massive experiments have been done on the training set and testing set in order to assess our refined fully convolutional neural networks and new types of loss function. Competitive figures prove they are more effective than their predecessors.
Measuring divergence between two distributions is essential in machine learning and statistics and has various applications including binary classification, change point detection, and two-sample test. Furthermore, in the era of big data, designing divergence measure that is interpretable and can handle high-dimensional and complex data becomes extremely important. In the paper, we propose a post selection inference (PSI) framework for divergence measure, which can select a set of statistically significant features that discriminate two distributions. Specifically, we employ an additive variant of maximum mean discrepancy (MMD) for features and introduce a general hypothesis test for PSI. A novel MMD estimator using the incomplete U-statistics, which has an asymptotically Normal distribution (under mild assumptions) and gives high detection power in PSI, is also proposed and analyzed theoretically. Through synthetic and real-world feature selection experiments, we show that the proposed framework can successfully detect statistically significant features. Last, we propose a sample selection framework for analyzing different members in the Generative Adversarial Networks (GANs) family.
Political speeches and debates play an important role in shaping the images of politicians, and the public often relies on media outlets to select bits of political communication from a large pool of utterances. It is an important research question to understand what factors impact this selection process. To quantitatively explore the selection process, we build a three- decade dataset of presidential debate transcripts and post-debate coverage. We first examine the effect of wording and propose a binary classification framework that controls for both the speaker and the debate situation. We find that crowdworkers can only achieve an accuracy of 60% in this task, indicating that media choices are not entirely obvious. Our classifiers outperform crowdworkers on average, mainly in primary debates. We also compare important factors from crowdworkers' free-form explanations with those from data-driven methods and find interesting differences. Few crowdworkers mentioned that "context matters", whereas our data show that well-quoted sentences are more distinct from the previous utterance by the same speaker than less-quoted sentences. Finally, we examine the aggregate effect of media preferences towards different wordings to understand the extent of fragmentation among media outlets. By analyzing a bipartite graph built from quoting behavior in our data, we observe a decreasing trend in bipartisan coverage.
Recent work has extended the theoretical analysis of boosting algorithms to multiclass problems and to online settings. However, the multiclass extension is in the batch setting and the online extensions only consider binary classification. We fill this gap in the literature by defining, and justifying, a weak learning condition for online multiclass boosting. This condition leads to an optimal boosting algorithm that requires the minimal number of weak learners to achieve a certain accuracy. Additionally, we propose an adaptive algorithm which is near optimal and enjoys an excellent performance on real data due to its adaptive property.
High-dimensional data requires scalable algorithms. We propose and analyze three scalable and related algorithms for semi-supervised discriminant analysis (SDA). These methods are based on Krylov subspace methods which exploit the data sparsity and the shift-invariance of Krylov subspaces. In addition, the problem definition was improved by adding centralization to the semi-supervised setting. The proposed methods are evaluated on a industry-scale data set from a pharmaceutical company to predict compound activity on target proteins. The results show that SDA achieves good predictive performance and our methods only require a few seconds, significantly improving computation time on previous state of the art.
We propose a novel online learning algorithm for Restricted Boltzmann Machines (RBM), namely, the Online Generative Discriminative Restricted Boltzmann Machine (OGD-RBM), that provides the ability to build and adapt the network architecture of RBM according to the statistics of streaming data. The OGD-RBM is trained in two phases: (1) an online generative phase for unsupervised feature representation at the hidden layer and (2) a discriminative phase for classification. The online generative training begins with zero neurons in the hidden layer, adds and updates the neurons to adapt to statistics of streaming data in a single pass unsupervised manner, resulting in a feature representation best suited to the data. The discriminative phase is based on stochastic gradient descent and associates the represented features to the class labels. We demonstrate the OGD-RBM on a set of multi-category and binary classification problems for data sets having varying degrees of class-imbalance. We first apply the OGD-RBM algorithm on the multi-class MNIST dataset to characterize the network evolution. We demonstrate that the online generative phase converges to a stable, concise network architecture, wherein individual neurons are inherently discriminative to the class labels despite unsupervised training. We then benchmark OGD-RBM performance to other machine learning, neural network and ClassRBM techniques for credit scoring applications using 3 public non-stationary two-class credit datasets with varying degrees of class-imbalance. We report that OGD-RBM improves accuracy by 2.5-3% over batch learning techniques while requiring at least 24%-70% fewer neurons and fewer training samples. This online generative training approach can be extended greedily to multiple layers for training Deep Belief Networks in non-stationary data mining applications without the need for a priori fixed architectures.
It is widely conjectured that the reason that training algorithms for neural networks are successful because all local minima lead to similar performance, for example, see (LeCun et al., 2015, Choromanska et al., 2015, Dauphin et al., 2014). Performance is typically measured in terms of two metrics: training performance and generalization performance. Here we focus on the training performance of single-layered neural networks for binary classification, and provide conditions under which the training error is zero at all local minima of a smooth hinge loss function. Our conditions are roughly in the following form: the neurons have to be strictly convex and the surrogate loss function should be a smooth version of hinge loss. We also provide counterexamples to show that when the loss function is replaced with quadratic loss or logistic loss, the result may not hold.
We present a systematic approach for achieving fairness in a binary classification setting. While we focus on two well-known quantitative definitions of fairness, our approach encompasses many other previously studied definitions as special cases. The key idea is to reduce fair classification to a sequence of cost-sensitive classification problems, whose solutions yield a randomized classifier with the lowest (empirical) error subject to the desired constraints. We introduce two reductions that work for any representation of the cost-sensitive classifier and compare favorably to prior baselines on a variety of data sets, while overcoming several of their disadvantages.
In this work, we present a globalized stochastic semismooth Newton method for solving stochastic optimization problems involving smooth nonconvex and nonsmooth convex terms in the objective function. We assume that only noisy gradient and Hessian information of the smooth part of the objective function is available via calling stochastic first and second order oracles. The proposed method can be seen as a hybrid approach combining stochastic semismooth Newton steps and stochastic proximal gradient steps. Two inexact growth conditions are incorporated to monitor the convergence and the acceptance of the semismooth Newton steps and it is shown that the algorithm converges globally to stationary points in expectation. Moreover, under standard assumptions and utilizing random matrix concentration inequalities, we prove that the proposed approach locally turns into a pure stochastic semismooth Newton method and converges r-superlinearly with high probability. We present numerical results and comparisons on $\ell_1$-regularized logistic regression and nonconvex binary classification that demonstrate the efficiency of our algorithm.
We present a new approach for mitigating unfairness in learned classifiers. In particular, we focus on binary classification tasks over individuals from two populations, where, as our criterion for fairness, we wish to achieve similar false positive rates in both populations, and similar false negative rates in both populations. As a proof of concept, we implement our approach and empirically evaluate its ability to achieve both fairness and accuracy, using datasets from the fields of criminal risk assessment, credit, lending, and college admissions.
We consider high-dimensional binary classification by sparse logistic regression. We propose a model/feature selection procedure based on penalized maximum likelihood with a complexity penalty on the model size and derive the non-asymptotic bounds for the resulting misclassification excess risk. The bounds can be reduced under the additional low-noise condition. The proposed complexity penalty is remarkably related to the VC-dimension of a set of sparse linear classifiers. Implementation of any complexity penalty-based criterion, however, requires a combinatorial search over all possible models. To find a model selection procedure computationally feasible for high-dimensional data, we extend the Slope estimator for logistic regression and show that under an additional weighted restricted eigenvalue condition it is rate-optimal in the minimax sense.
Gated Recurrent Unit (GRU) is a recently-developed variation of the long short-term memory (LSTM) unit, both of which are types of recurrent neural network (RNN). Through empirical evidence, both models have been proven to be effective in a wide variety of machine learning tasks such as natural language processing (Wen et al., 2015), speech recognition (Chorowski et al., 2015), and text classification (Yang et al., 2016). Conventionally, like most neural networks, both of the aforementioned RNN variants employ the Softmax function as its final output layer for its prediction, and the cross-entropy function for computing its loss. In this paper, we present an amendment to this norm by introducing linear support vector machine (SVM) as the replacement for Softmax in the final output layer of a GRU model. Furthermore, the cross-entropy function shall be replaced with a margin-based function. While there have been similar studies (Alalshekmubarak & Smith, 2013; Tang, 2013), this proposal is primarily intended for binary classification on intrusion detection using the 2013 network traffic data from the honeypot systems of Kyoto University. Results show that the GRU-SVM model performs relatively higher than the conventional GRU-Softmax model. The proposed model reached a training accuracy of ~81.54% and a testing accuracy of ~84.15%, while the latter was able to reach a training accuracy of ~63.07% and a testing accuracy of ~70.75%. In addition, the juxtaposition of these two final output layers indicate that the SVM would outperform Softmax in prediction time - a theoretical implication which was supported by the actual training and testing time in the study.
We design differentially private learning algorithms that are agnostic to the learning model. Our algorithms are interactive in nature, i.e., instead of outputting a model based on the training data, they provide predictions for a set of $m$ feature vectors that arrive online. We show that, for the feature vectors on which an ensemble of models (trained on random disjoint subsets of a dataset) makes consistent predictions, there is almost no-cost of privacy in generating accurate predictions for those feature vectors. To that end, we provide a novel coupling of the distance to instability framework with the sparse vector technique. We provide algorithms with formal privacy and utility guarantees for both binary/multi-class classification, and soft-label classification. For binary classification in the standard (agnostic) PAC model, we show how to bootstrap from our privately generated predictions to construct a computationally efficient private learner that outputs a final accurate hypothesis. Our construction - to the best of our knowledge - is the first computationally efficient construction for a label-private learner. We prove sample complexity upper bounds for this setting. As in non-private sample complexity bounds, the only relevant property of the given concept class is its VC dimension. For soft-label classification, our techniques are based on exploiting the stability properties of traditional learning algorithms, like stochastic gradient descent (SGD). We provide a new technique to boost the average-case stability properties of learning algorithms to strong (worst-case) stability properties, and then exploit them to obtain private classification algorithms. In the process, we also show that a large class of SGD methods satisfy average-case stability properties, in contrast to a smaller class of SGD methods that are uniformly stable as shown in prior work.
A falling rule list is a probabilistic decision list for binary classification, consisting of a series of if-then rules with antecedents in the if clauses and probabilities of the desired outcome ("1") in the then clauses. Just as in a regular decision list, the order of rules in a falling rule list is important -- each example is classified by the first rule whose antecedent it satisfies. Unlike a regular decision list, a falling rule list requires the probabilities of the desired outcome ("1") to be monotonically decreasing down the list. We propose an optimization approach to learning falling rule lists and "softly" falling rule lists, along with Monte-Carlo search algorithms that use bounds on the optimal solution to prune the search space.
We present a new approach to evaluate computational models for the task of text understanding by the means of out-of-context error detection. Through the novel design of our automated modification process, existing large-scale data sources can be adopted for a vast number of text understanding tasks. The data is thereby altered on a semantic level, allowing models to be tested against a challenging set of modified text passages that require to comprise a broader narrative discourse. Our newly introduced task targets actual real-world problems of transcription and translation systems by inserting authentic out-of-context errors. The automated modification process is applied to the 2016 TEDTalk corpus. Entirely automating the process allows the adoption of complete datasets at low cost, facilitating supervised learning procedures and deeper networks to be trained and tested. To evaluate the quality of the modification algorithm a language model and a supervised binary classification model are trained and tested on the altered dataset. A human baseline evaluation is examined to compare the results with human performance. The outcome of the evaluation task indicates the difficulty to detect semantic errors for machine-learning algorithms and humans, showing that the errors cannot be identified when limited to a single sentence.
Optimization of patient throughput and wait time in emergency departments (ED) is an important task for hospital systems. For that reason, Emergency Severity Index (ESI) system for patient triage was introduced to help guide manual estimation of acuity levels, which is used by nurses to rank the patients and organize hospital resources. However, despite improvements that it brought to managing medical resources, such triage system greatly depends on nurse's subjective judgment and is thus prone to human errors. Here, we propose a novel deep model based on the word attention mechanism designed for predicting a number of resources an ED patient would need. Our approach incorporates routinely available continuous and nominal (structured) data with medical text (unstructured) data, including patient's chief complaint, past medical history, medication list, and nurse assessment collected for 338,500 ED visits over three years in a large urban hospital. Using both structured and unstructured data, the proposed approach achieves the AUC of $\sim 88\%$ for the task of identifying resource intensive patients (binary classification), and the accuracy of $\sim 44\%$ for predicting exact category of number of resources (multi-class classification task), giving an estimated lift over nurses' performance by 16\% in accuracy. Furthermore, the attention mechanism of the proposed model provides interpretability by assigning attention scores for nurses' notes which is crucial for decision making and implementation of such approaches in the real systems working on human health.
Face anti-spoofing is the crucial step to prevent face recognition systems from a security breach. Previous deep learning approaches formulate face anti-spoofing as a binary classification problem. Many of them struggle to grasp adequate spoofing cues and generalize poorly. In this paper, we argue the importance of auxiliary supervision to guide the learning toward discriminative and generalizable cues. A CNN-RNN model is learned to estimate the face depth with pixel-wise supervision, and to estimate rPPG signals with sequence-wise supervision. Then we fuse the estimated depth and rPPG to distinguish live vs. spoof faces. In addition, we introduce a new face anti-spoofing database that covers a large range of illumination, subject, and pose variations. Experimental results show that our model achieves the state-of-the-art performance on both intra-database and cross-database testing.
For information retrieval and binary classification, we show that precision at the top (or precision at k) and recall at the top (or recall at k) are maximised by thresholding the posterior probability of the positive class. This finding is a consequence of a result on constrained minimisation of the cost-sensitive expected classification error which generalises an earlier related result from the literature.