We propose Very Simple Classifier (VSC) a novel method designed to incorporate the concepts of subsampling and locality in the definition of features to be used as the input of a perceptron. The rationale is that locality theoretically guarantees a bound on the generalization error. Each feature in VSC is a max-margin classifier built on randomly-selected pairs of samples. The locality in VSC is achieved by multiplying the value of the feature by a confidence measure that can be characterized in terms of the Chebichev inequality. The output of the layer is then fed in a output layer of neurons. The weights of the output layer are then determined by a regularized pseudoinverse. Extensive comparison of VSC against 9 competitors in the task of binary classification is carried out. Results on 22 benchmark datasets with fixed parameters show that VSC is competitive with the Multi Layer Perceptron (MLP) and outperforms the other competitors. An exploration of the parameter space shows VSC can outperform MLP.
In binary classification and regression problems, it is well understood that Lipschitz continuity and smoothness of the loss function play key roles in governing generalization error bounds for empirical risk minimization algorithms. In this paper, we show how these two properties affect generalization error bounds in the learning to rank problem. The learning to rank problem involves vector valued predictions and therefore the choice of the norm with respect to which Lipschitz continuity and smoothness are defined becomes crucial. Choosing the $\ell_\infty$ norm in our definition of Lipschitz continuity allows us to improve existing bounds. Furthermore, under smoothness assumptions, our choice enables us to prove rates that interpolate between $1/\sqrt{n}$ and $1/n$ rates. Application of our results to ListNet, a popular learning to rank method, gives state-of-the-art performance guarantees.
Ensemble methods for classification and clustering have been effectively used for decades, while ensemble learning for outlier detection has only been studied recently. In this work, we design a new ensemble approach for outlier detection in multi-dimensional point data, which provides improved accuracy by reducing error through both bias and variance. Although classification and outlier detection appear as different problems, their theoretical underpinnings are quite similar in terms of the bias-variance trade-off [1], where outlier detection is considered as a binary classification task with unobserved labels but a similar bias-variance decomposition of error. In this paper, we propose a sequential ensemble approach called CARE that employs a two-phase aggregation of the intermediate results in each iteration to reach the final outcome. Unlike existing outlier ensembles which solely incorporate a parallel framework by aggregating the outcomes of independent base detectors to reduce variance, our ensemble incorporates both the parallel and sequential building blocks to reduce bias as well as variance by ($i$) successively eliminating outliers from the original dataset to build a better data model on which outlierness is estimated (sequentially), and ($ii$) combining the results from individual base detectors and across iterations (parallelly). Through extensive experiments on sixteen real-world datasets mainly from the UCI machine learning repository [2], we show that CARE performs significantly better than or at least similar to the individual baselines. We also compare CARE with the state-of-the-art outlier ensembles where it also provides significant improvement when it is the winner and remains close otherwise.
A methodology for binary classification of EEG records which correspond to different mental states is proposed. This model-free methodology is based on our theory of the $\epsilon$-complexity of continuous functions which is extended here (see Appendix) to the case of vector functions. This extension permits us to handle multichannel EEG recordings. The essence of the methodology is to use the $\epsilon$-complexity coefficients as features to classify (using well known classifiers) different types of vector functions representing EEG-records corresponding to different types of mental states. We apply our methodology to the problem of classification of multichannel EEG-records related to a group of healthy adolescents and a group of adolescents with schizophrenia. We found that our methodology permits accurate classification of the data in the four-dimensional feather space of the $\epsilon$-complexity coefficients.
We consider optimization of generalized performance metrics for binary classification by means of surrogate losses. We focus on a class of metrics, which are linear-fractional functions of the false positive and false negative rates (examples of which include $F_{\beta}$-measure, Jaccard similarity coefficient, AM measure, and many others). Our analysis concerns the following two-step procedure. First, a real-valued function $f$ is learned by minimizing a surrogate loss for binary classification on the training sample. It is assumed that the surrogate loss is a strongly proper composite loss function (examples of which include logistic loss, squared-error loss, exponential loss, etc.). Then, given $f$, a threshold $\widehat{\theta}$ is tuned on a separate validation sample, by direct optimization of the target performance metric. We show that the regret of the resulting classifier (obtained from thresholding $f$ on $\widehat{\theta}$) measured with respect to the target metric is upperbounded by the regret of $f$ measured with respect to the surrogate loss. We also extend our results to cover multilabel classification and provide regret bounds for micro- and macro-averaging measures. Our findings are further analyzed in a computational study on both synthetic and real data sets.
A particularly interesting instance of supervised learning with kernels is when each training example is associated with two objects, as in pairwise classification (Brunner et al., 2012), and in supervised learning of preference relations (Herbrich et al., 1998). In these cases, one may want to embed additional prior knowledge into the optimization problem associated with the training of the learning machine, modeled, respectively, by the symmetry of its optimal solution with respect to an exchange of order between the two objects, and by its antisymmetry. Extending the approach proposed in (Brunner et al., 2012) (where the only symmetric case was considered), we show, focusing on support vector binary classification, how such embedding is possible through the choice of a suitable pairwise kernel, which takes as inputs the individual feature vectors and also the group feature vectors associated with the two objects. We also prove that the symmetry/antisymmetry constraints still hold when considering the sequence of suboptimal solutions generated by one version of the Sequential Minimal Optimization (SMO) algorithm, and we present numerical results supporting the theoretical findings. We conclude discussing extensions of the main results to support vector regression, to transductive support vector machines, and to several kinds of graph kernels, including diffusion kernels.
We analyse an issue when comparing survival curves between two subgroups. We show that there is a direct relationship between estimates of subgroups' survival at a time point and positive and negative predictive values in the binary classification settings. Our findings present a case where current methods of comparing survival curves between subgroups may be misleading. We think that this ought to be taken into account during the validation of prognostic diagnostic tests that predict two prognostic subgroups for a given disease or treatment, when the validation data set consists of censored data.
Various families of malware use domain generation algorithms (DGAs) to generate a large number of pseudo-random domain names to connect to a command and control (C&C) server. In order to block DGA C&C traffic, security organizations must first discover the algorithm by reverse engineering malware samples, then generating a list of domains for a given seed. The domains are then either preregistered or published in a DNS blacklist. This process is not only tedious, but can be readily circumvented by malware authors using a large number of seeds in algorithms with multivariate recurrence properties (e.g., banjori) or by using a dynamic list of seeds (e.g., bedep). Another technique to stop malware from using DGAs is to intercept DNS queries on a network and predict whether domains are DGA generated. Such a technique will alert network administrators to the presence of malware on their networks. In addition, if the predictor can also accurately predict the family of DGAs, then network administrators can also be alerted to the type of malware that is on their networks. This paper presents a DGA classifier that leverages long short-term memory (LSTM) networks to predict DGAs and their respective families without the need for a priori feature extraction. Results are significantly better than state-of-the-art techniques, providing 0.9993 area under the receiver operating characteristic curve for binary classification and a micro-averaged F1 score of 0.9906. In other terms, the LSTM technique can provide a 90% detection rate with a 1:10000 false positive (FP) rate---a twenty times FP improvement over comparable methods. Experiments in this paper are run on open datasets and code snippets are provided to reproduce the results.
We address the problem of aggregating an ensemble of predictors with known loss bounds in a semi-supervised binary classification setting, to minimize prediction loss incurred on the unlabeled data. We find the minimax optimal predictions for a very general class of loss functions including all convex and many non-convex losses, extending a recent analysis of the problem for misclassification error. The result is a family of semi-supervised ensemble aggregation algorithms which are as efficient as linear learning by convex optimization, but are minimax optimal without any relaxations. Their decision rules take a form familiar in decision theory -- applying sigmoid functions to a notion of ensemble margin -- without the assumptions typically made in margin-based learning.
In stochastic optimization, the population risk is generally approximated by the empirical risk. However, in the large-scale setting, minimization of the empirical risk may be computationally restrictive. In this paper, we design an efficient algorithm to approximate the population risk minimizer in generalized linear problems such as binary classification with surrogate losses and generalized linear regression models. We focus on large-scale problems, where the iterative minimization of the empirical risk is computationally intractable, i.e., the number of observations $n$ is much larger than the dimension of the parameter $p$, i.e. $n \gg p \gg 1$. We show that under random sub-Gaussian design, the true minimizer of the population risk is approximately proportional to the corresponding ordinary least squares (OLS) estimator. Using this relation, we design an algorithm that achieves the same accuracy as the empirical risk minimizer through iterations that attain up to a cubic convergence rate, and that are cheaper than any batch optimization algorithm by at least a factor of $\mathcal{O}(p)$. We provide theoretical guarantees for our algorithm, and analyze the convergence behavior in terms of data dimensions. Finally, we demonstrate the performance of our algorithm on well-known classification and regression problems, through extensive numerical studies on large-scale datasets, and show that it achieves the highest performance compared to several other widely used and specialized optimization algorithms.
Recurrent Neural Network (RNN) is one of the most popular architectures used in Natural Language Processsing (NLP) tasks because its recurrent structure is very suitable to process variable-length text. RNN can utilize distributed representations of words by first converting the tokens comprising each text into vectors, which form a matrix. And this matrix includes two dimensions: the time-step dimension and the feature vector dimension. Then most existing models usually utilize one-dimensional (1D) max pooling operation or attention-based operation only on the time-step dimension to obtain a fixed-length vector. However, the features on the feature vector dimension are not mutually independent, and simply applying 1D pooling operation over the time-step dimension independently may destroy the structure of the feature representation. On the other hand, applying two-dimensional (2D) pooling operation over the two dimensions may sample more meaningful features for sequence modeling tasks. To integrate the features on both dimensions of the matrix, this paper explores applying 2D max pooling operation to obtain a fixed-length representation of the text. This paper also utilizes 2D convolution to sample more meaningful information of the matrix. Experiments are conducted on six text classification tasks, including sentiment analysis, question classification, subjectivity classification and newsgroup classification. Compared with the state-of-the-art models, the proposed models achieve excellent performance on 4 out of 6 tasks. Specifically, one of the proposed models achieves highest accuracy on Stanford Sentiment Treebank binary classification and fine-grained classification tasks.
Visual question answering (VQA) is an interesting learning setting for evaluating the abilities and shortcomings of current systems for image understanding. Many of the recently proposed VQA systems include attention or memory mechanisms designed to support "reasoning". For multiple-choice VQA, nearly all of these systems train a multi-class classifier on image and question features to predict an answer. This paper questions the value of these common practices and develops a simple alternative model based on binary classification. Instead of treating answers as competing choices, our model receives the answer as input and predicts whether or not an image-question-answer triplet is correct. We evaluate our model on the Visual7W Telling and the VQA Real Multiple Choice tasks, and find that even simple versions of our model perform competitively. Our best model achieves state-of-the-art performance on the Visual7W Telling task and compares surprisingly well with the most complex systems proposed for the VQA Real Multiple Choice task. We explore variants of the model and study its transferability between both datasets. We also present an error analysis of our model that suggests a key problem of current VQA systems lies in the lack of visual grounding of concepts that occur in the questions and answers. Overall, our results suggest that the performance of current VQA systems is not significantly better than that of systems designed to exploit dataset biases.
We consider a large dataset of real-world, on-road driving from a 100-car naturalistic study to explore the predictive power of driver glances and, specifically, to answer the following question: what can be predicted about the state of the driver and the state of the driving environment from a 6-second sequence of macro-glances? The context-based nature of such glances allows for application of supervised learning to the problem of vision-based gaze estimation, making it robust, accurate, and reliable in messy, real-world conditions. So, it's valuable to ask whether such macro-glances can be used to infer behavioral, environmental, and demographic variables? We analyze 27 binary classification problems based on these variables. The takeaway is that glance can be used as part of a multi-sensor real-time system to predict radio-tuning, fatigue state, failure to signal, talking, and several environment variables.
Person re-identification (ReID) focuses on identifying people across different scenes in video surveillance, which is usually formulated as a binary classification task or a ranking task in current person ReID approaches. In this paper, we take both tasks into account and propose a multi-task deep network (MTDnet) that makes use of their own advantages and jointly optimize the two tasks simultaneously for person ReID. To the best of our knowledge, we are the first to integrate both tasks in one network to solve the person ReID. We show that our proposed architecture significantly boosts the performance. Furthermore, deep architecture in general requires a sufficient dataset for training, which is usually not met in person ReID. To cope with this situation, we further extend the MTDnet and propose a cross-domain architecture that is capable of using an auxiliary set to assist training on small target sets. In the experiments, our approach outperforms most of existing person ReID algorithms on representative datasets including CUHK03, CUHK01, VIPeR, iLIDS and PRID2011, which clearly demonstrates the effectiveness of the proposed approach.
Identifying and correcting grammatical errors in the text written by non-native writers has received increasing attention in recent years. Although a number of annotated corpora have been established to facilitate data-driven grammatical error detection and correction approaches, they are still limited in terms of quantity and coverage because human annotation is labor-intensive, time-consuming, and expensive. In this work, we propose to utilize unlabeled data to train neural network based grammatical error detection models. The basic idea is to cast error detection as a binary classification problem and derive positive and negative training examples from unlabeled data. We introduce an attention-based neural network to capture long-distance dependencies that influence the word being detected. Experiments show that the proposed approach significantly outperforms SVMs and convolutional networks with fixed-size context window.
Multiple Sclerosis is a degenerative condition of the central nervous system that affects nearly 2.5 million of individuals in terms of their physical, cognitive, psychological and social capabilities. Researchers are currently investigating on the use of patient reported outcome measures for the assessment of impact and evolution of the disease on the life of the patients. To date, a clear understanding on the use of such measures to predict the evolution of the disease is still lacking. In this work we resort to regularized machine learning methods for binary classification and multiple output regression. We propose a pipeline that can be used to predict the disease progression from patient reported measures. The obtained model is tested on a data set collected from an ongoing clinical research project.
We create a new online reduction of multiclass classification to binary classification for which training and prediction time scale logarithmically with the number of classes. Compared to previous approaches, we obtain substantially better statistical performance for two reasons: First, we prove a tighter and more complete boosting theorem, and second we translate the results more directly into an algorithm. We show that several simple techniques give rise to an algorithm that can compete with one-against-all in both space and predictive power while offering exponential improvements in speed when the number of classes is large.
We introduce several techniques for sampling and visualizing the latent spaces of generative models. Replacing linear interpolation with spherical linear interpolation prevents diverging from a model's prior distribution and produces sharper samples. J-Diagrams and MINE grids are introduced as visualizations of manifolds created by analogies and nearest neighbors. We demonstrate two new techniques for deriving attribute vectors: bias-corrected vectors with data replication and synthetic vectors with data augmentation. Binary classification using attribute vectors is presented as a technique supporting quantitative analysis of the latent space. Most techniques are intended to be independent of model type and examples are shown on both Variational Autoencoders and Generative Adversarial Networks.
In this paper, we propose two methods for tackling the problem of cross-device matching for online advertising at CIKM Cup 2016. The first method considers the matching problem as a binary classification task and solve it by utilizing ensemble learning techniques. The second method defines the matching problem as a ranking task and effectively solve it with using learning-to-rank algorithms. The results show that the proposed methods obtain promising results, in which the ranking-based method outperforms the classification-based method for the task.
This work continues the study of the relationship between sample compression schemes and statistical learning, which has been mostly investigated within the framework of binary classification. The central theme of this work is establishing equivalences between learnability and compressibility, and utilizing these equivalences in the study of statistical learning theory. We begin with the setting of multiclass categorization (zero/one loss). We prove that in this case learnability is equivalent to compression of logarithmic sample size, and that uniform convergence implies compression of constant size. We then consider Vapnik's general learning setting: we show that in order to extend the compressibility-learnability equivalence to this case, it is necessary to consider an approximate variant of compression. Finally, we provide some applications of the compressibility-learnability equivalences: (i) Agnostic-case learnability and realizable-case learnability are equivalent in multiclass categorization problems (in terms of sample complexity). (ii) This equivalence between agnostic-case learnability and realizable-case learnability does not hold for general learning problems: There exists a learning problem whose loss function takes just three values, under which agnostic-case and realizable-case learnability are not equivalent. (iii) Uniform convergence implies compression of constant size in multiclass categorization problems. Part of the argument includes an analysis of the uniform convergence rate in terms of the graph dimension, in which we improve upon previous bounds. (iv) A dichotomy for sample compression in multiclass categorization problems: If a non-trivial compression exists then a compression of logarithmic size exists. (v) A compactness theorem for multiclass categorization problems.