In supervised machine learning, the assumption that training data is labelled correctly is not always satisfied. In this paper, we investigate an instance of labelling error for classification tasks in which the dataset is corrupted with out-of-distribution (OOD) instances: data that does not belong to any of the target classes, but is labelled as such. We show that detecting and relabelling certain OOD instances, rather than discarding them, can have a positive effect on learning. The proposed method uses an auxiliary classifier, trained on data that is known to be in-distribution, for detection and relabelling. The amount of data required for this is shown to be small. Experiments are carried out on the FSDnoisy18k audio dataset, where OOD instances are very prevalent. The proposed method is shown to improve the performance of convolutional neural networks by a significant margin. Comparisons with other noise-robust techniques are similarly encouraging.
A novel correction algorithm is proposed for multi-class classification problems with corrupted training data. The algorithm is non-intrusive, in the sense that it post-processes a trained classification model by adding a correction procedure to the model prediction. The correction procedure can be coupled with any approximators, such as logistic regression, neural networks of various architectures, etc. When training dataset is sufficiently large, we prove that the corrected models deliver correct classification results as if there is no corruption in the training data. For datasets of finite size, the corrected models produce significantly better recovery results, compared to the models without the correction algorithm. All of the theoretical findings in the paper are verified by our numerical examples.
Recent work on policy learning from observational data has highlighted the importance of efficient policy evaluation and has proposed reductions to weighted (cost-sensitive) classification. But, efficient policy evaluation need not yield efficient estimation of policy parameters. We consider the estimation problem given by a weighted surrogate-loss classification reduction of policy learning with any score function, either direct, inverse-propensity weighted, or doubly robust. We show that, under a correct specification assumption, the weighted classification formulation need not be efficient for policy parameters. We draw a contrast to actual (possibly weighted) binary classification, where correct specification implies a parametric model, while for policy learning it only implies a semiparametric model. In light of this, we instead propose an estimation approach based on generalized method of moments, which is efficient for the policy parameters. We propose a particular method based on recent developments on solving moment problems using neural networks and demonstrate the efficiency and regret benefits of this method empirically.
Deep neural networks for video classification, just like image classification networks, may be subjected to adversarial manipulation. The main difference between image classifiers and video classifiers is that the latter usually use temporal information contained within the video. In this work we present a manipulation scheme for fooling video classifiers by introducing a flickering temporal perturbation that in some cases may be unnoticeable by human observers and is implementable in the real world. After demonstrating the manipulation of action classification of single videos, we generalize the procedure to make universal adversarial perturbation, achieving high fooling ratio. In addition, we generalize the universal perturbation and produce a temporal-invariant perturbation, which can be applied to the video without synchronizing the perturbation to the input. The attack was implemented on several target models and the transferability of the attack was demonstrated. These properties allow us to bridge the gap between simulated environment and real-world application, as will be demonstrated in this paper for the first time for an over-the-air flickering attack.
An increasing amount of collected data are high-dimensional multi-way arrays (tensors), and it is crucial for efficient learning algorithms to exploit this tensorial structure as much as possible. The ever-present curse of dimensionality for high dimensional data and the loss of structure when vectorizing the data motivates the use of tailored low-rank tensor classification methods. In the presence of small amounts of training data, kernel methods offer an attractive choice as they provide the possibility for a nonlinear decision boundary. We develop the Tensor Train Multi-way Multi-level Kernel (TT-MMK), which combines the simplicity of the Canonical Polyadic decomposition, the classification power of the Dual Structure-preserving Support Vector Machine, and the reliability of the Tensor Train (TT) approximation. We show by experiments that the TT-MMK method is usually more reliable computationally, less sensitive to tuning parameters, and gives higher prediction accuracy in the SVM classification when benchmarked against other state-of-the-art techniques.
Relationships in online social networks often imply social connections in the real world. An accurate understanding of relationship types benefits many applications, e.g. social advertising and recommendation. Some recent attempts have been proposed to classify user relationships into predefined types with the help of pre-labeled relationships or abundant interaction features on relationships. Unfortunately, both relationship feature data and label data are very sparse in real social platforms like WeChat, rendering existing methods inapplicable. In this paper, we present an in-depth analysis of WeChat relationships to identify the major challenges for the relationship classification task. To tackle the challenges, we propose a Local Community-based Edge Classification (LoCEC) framework that classifies user relationships in a social network into real-world social connection types. LoCEC enforces a three-phase processing, namely local community detection, community classification and relationship classification, to address the sparsity issue of relationship features and relationship labels. Moreover, LoCEC is designed to handle large-scale networks by allowing parallel and distributed processing. We conduct extensive experiments on the real-world WeChat network with hundreds of billions of edges to validate the effectiveness and efficiency of LoCEC.
A common issue for classification in scientific research and industry is the existence of imbalanced classes. When sample sizes of different classes are imbalanced in training data, naively implementing a classification method often leads to unsatisfactory prediction results on test data. Multiple resampling techniques have been proposed to address the class imbalance issues. Yet, there is no general guidance on when to use each technique. In this article, we provide a paradigm-based review of the common resampling techniques for binary classification under imbalanced class sizes. The paradigms we consider include the classical paradigm that minimizes the overall classification error, the cost-sensitive learning paradigm that minimizes a cost-adjusted weighted type I and type II errors, and the Neyman-Pearson paradigm that minimizes the type II error subject to a type I error constraint. Under each paradigm, we investigate the combination of the resampling techniques and a few state-of-the-art classification methods. For each pair of resampling techniques and classification methods, we use simulation studies and a real data set on credit card fraud to study the performance under different evaluation metrics. From these extensive numerical experiments, we demonstrate under each classification paradigm, the complex dynamics among resampling techniques, base classification methods, evaluation metrics, and imbalance ratios. We also summarize a few takeaway messages regarding the choices of resampling techniques and base classification methods, which could be helpful for practitioners.
We present an approach to explain the decisions of black box models for image classification. While using the black box to label images, our explanation method exploits the latent feature space learned through an adversarial autoencoder. The proposed method first generates exemplar images in the latent feature space and learns a decision tree classifier. Then, it selects and decodes exemplars respecting local decision rules. Finally, it visualizes them in a manner that shows to the user how the exemplars can be modified to either stay within their class, or to become counter-factuals by "morphing" into another class. Since we focus on black box decision systems for image classification, the explanation obtained from the exemplars also provides a saliency map highlighting the areas of the image that contribute to its classification, and areas of the image that push it into another class. We present the results of an experimental evaluation on three datasets and two black box models. Besides providing the most useful and interpretable explanations, we show that the proposed method outperforms existing explainers in terms of fidelity, relevance, coherence, and stability.
Image classification requires the generation of features capable of detecting image patterns informative of group identity. The objective of this study was to classify images from the public CIFAR-10 image dataset by leveraging combinations of disparate image feature sources from both manual and deep learning approaches. Histogram of oriented gradients (HOG) and pixel intensities successfully inform classification (53% and 59% classification accuracy, respectively), yet there is much room for improvement. VGG16 with ImageNet trained weights and a CIFAR-10 optimized model (CIFAR-VGG) further improve upon image classification (60% and 93.43% accuracy, respectively). We further improved classification by utilizing transfer learning to re-establish optimal network weights for VGG16 (TL-VGG) and Inception ResNet v2 (TL-Inception) resulting in significant performance increases (85% and 90.74%, respectively), yet fail to surpass CIFAR-VGG. We hypothesized that if each generated feature set obtained some unique insight into the classification problem, then combining these features would result in greater classification accuracy, surpassing that of CIFAR-VGG. Upon selection of the top 1000 principal components from TL-VGG, TL-Inception, HOG, pixel intensities, and CIFAR-VGG, we achieved testing accuracy of 94.6%, lending support to our hypothesis.
We consider the problems of classification and intrinsic dimension estimation on image data. A new subspace based classifier is proposed for supervised classification or intrinsic dimension estimation. The distribution of the data in each class is modeled by a union of of a finite number ofaffine subspaces of the feature space. The affine subspaces have a common dimension, which is assumed to be much less than the dimension of the feature space. The subspaces are found using regression based on the L0-norm. The proposed method is a generalisation of classical NN (Nearest Neighbor), NFL (Nearest Feature Line) classifiers and has a close relationship to NS (Nearest Subspace) classifier. The proposed classifier with an accurately estimated dimension parameter generally outperforms its competitors in terms of classification accuracy. We also propose a fast version of the classifier using a neighborhood representation to reduce its computational complexity. Experiments on publicly available datasets corroborate these claims.
Overfitting and treatment of "small data" are among the most challenging problems in the machine learning (ML), when a relatively small data statistics size $T$ is not enough to provide a robust ML fit for a relatively large data feature dimension $D$. Deploying a massively-parallel ML analysis of generic classification problems for different $D$ and $T$, existence of statistically-significant linear overfitting barriers for common ML methods is demonstrated. For example, these results reveal that for a robust classification of bioinformatics-motivated generic problems with the Long Short-Term Memory deep learning classifier (LSTM) one needs in a best case a statistics $T$ that is at least 13.8 times larger then the feature dimension $D$. It is shown that this overfitting barrier can be breached at a $10^{-12}$ fraction of the computational cost by means of the entropy-optimal Scalable Probabilistic Approximations algorithm (eSPA), performing a joint solution of the entropy-optimal Bayesian network inference and feature space segmentation problems. Application of eSPA to experimental single cell RNA sequencing data exhibits a 30-fold classification performance boost when compared to standard bioinformatics tools - and a 7-fold boost when compared to the deep learning LSTM classifier.
The successive subspace learning (SSL) principle was developed and used to design an interpretable learning model, known as the PixelHop method,for image classification in our prior work. Here, we propose an improved PixelHop method and call it PixelHop++. First, to make the PixelHop model size smaller, we decouple a joint spatial-spectral input tensor to multiple spatial tensors (one for each spectral component) under the spatial-spectral separability assumption and perform the Saab transform in a channel-wise manner, called the channel-wise (c/w) Saab transform.Second, by performing this operation from one hop to another successively, we construct a channel-decomposed feature tree whose leaf nodes contain features of one dimension (1D). Third, these 1D features are ranked according to their cross-entropy values, which allows us to select a subset of discriminant features for image classification. In PixelHop++, one can control the learning model size of fine-granularity,offering a flexible tradeoff between the model size and the classification performance. We demonstrate the flexibility of PixelHop++ on MNIST, Fashion MNIST, and CIFAR-10 three datasets.
The task of text classification is usually divided into two stages: {\it text feature extraction} and {\it classification}. In this standard formalization categories are merely represented as indexes in the label vocabulary, and the model lacks for explicit instructions on what to classify. Inspired by the current trend of formalizing NLP problems as question answering tasks, we propose a new framework for text classification, in which each category label is associated with a category description. Descriptions are generated by hand-crafted templates or using abstractive/extractive models from reinforcement learning. The concatenation of the description and the text is fed to the classifier to decide whether or not the current label should be assigned to the text. The proposed strategy forces the model to attend to the most salient texts with respect to the label, which can be regarded as a hard version of attention, leading to better performances. We observe significant performance boosts over strong baselines on a wide range of text classification tasks including single-label classification, multi-label classification and multi-aspect sentiment analysis.
Modern AI image classifiers have made impressive advances in recent years, but their performance often appears strange or violates expectations of users. This suggests humans engage in cognitive anthropomorphism: expecting AI to have the same nature as human intelligence. This mismatch presents an obstacle to appropriate human-AI interaction. To delineate this mismatch, I examine known properties of human classification, in comparison to image classifier systems. Based on this examination, I offer three strategies for system design that can address the mismatch between human and AI classification: explainable AI, novel methods for training users, and new algorithms that match human cognition.
The 01 loss is robust to outliers and tolerant to noisy data compared to convex loss functions. We conjecture that the 01 loss may also be more robust to adversarial attacks. To study this empirically we have developed a stochastic coordinate descent algorithm for a linear 01 loss classifier and a single hidden layer 01 loss neural network. Due to the absence of the gradient we iteratively update coordinates on random subsets of the data for fixed epochs. We show our algorithms to be fast and comparable in accuracy to the linear support vector machine and logistic loss single hidden layer network for binary classification on several image benchmarks, thus establishing that our method is on-par in test accuracy with convex losses. We then subject them to accurately trained substitute model black box attacks on the same image benchmarks and find them to be more robust than convex counterparts. On CIFAR10 binary classification task between classes 0 and 1 with adversarial perturbation of 0.0625 we see that the MLP01 network loses 27\% in accuracy whereas the MLP-logistic counterpart loses 83\%. Similarly on STL10 and ImageNet binary classification between classes 0 and 1 the MLP01 network loses 21\% and 20\% while MLP-logistic loses 67\% and 45\% respectively. On MNIST that is a well-separable dataset we find MLP01 comparable to MLP-logistic and show under simulation how and why our 01 loss solver is less robust there. We then propose adversarial training for our linear 01 loss solver that significantly improves its robustness on MNIST and all other datasets and retains clean test accuracy. Finally we show practical applications of our method to deter traffic sign and facial recognition adversarial attacks. We discuss attacks with 01 loss, substitute model accuracy, and several future avenues like multiclass, 01 loss convolutions, and further adversarial training.
Recent years have witnessed the success of dictionary learning (DL) based approaches in the domain of pattern classification. In this paper, we present an efficient structured dictionary learning (ESDL) method which takes both the diversity and label information of training samples into account. Specifically, ESDL introduces alternative training samples into the process of dictionary learning. To increase the discriminative capability of representation coefficients for classification, an ideal regularization term is incorporated into the objective function of ESDL. Moreover, in contrast with conventional DL approaches which impose computationally expensive L1-norm constraint on the coefficient matrix, ESDL employs L2-norm regularization term. Experimental results on benchmark databases (including four face databases and one scene dataset) demonstrate that ESDL outperforms previous DL approaches. More importantly, ESDL can be applied in a wide range of pattern classification tasks.
Image data has been greatly produced by individuals and commercial vendors in the daily life, and it has been used across various domains, like advertising, medical and traffic analysis. Recently, image data also appears to be greatly important in social utility, like emergency response. However, the privacy concern becomes the biggest obstacle that prevents further exploration of image data, due to that the image could reveal sensitive information, like the personal identity and locations. The recent developed Local Differential Privacy (LDP) brings us a promising solution, which allows the data owners to randomly perturb their input to provide the plausible deniability of the data before releasing. In this paper, we consider a two-party image classification problem, in which data owners hold the image and the untrustworthy data user would like to fit a machine learning model with these images as input. To protect the image privacy, we propose to locally perturb the image representation before revealing to the data user. Subsequently, we analyze how the perturbation satisfies {\epsilon}-LDP and affect the data utility regarding count-based and distance-based machine learning algorithm, and propose a supervised image feature extractor, DCAConv, which produces an image representation with scalable domain size. Our experiments show that DCAConv could maintain a high data utility while preserving the privacy regarding multiple image benchmark datasets.
Cluttering of SMS inbox is one of the serious problems that users today face in the digital world where every online login, transaction, along with promotions generate multiple SMS. This problem not only prevents users from searching and navigating messages efficiently but often results in users missing out the relevant information associated with the corresponding SMS like offer codes, payment reminders etc. In this paper, we propose a unique architecture to organize and extract the appropriate information from SMS and further display it in an intuitive template. In the proposed architecture, we use a Hybrid Hierarchical Long Short Term Memory (LSTM)-Convolutional Neural Network (CNN) to categorize SMS into multiple classes followed by a set of entity parsers used to extract the relevant information from the classified message. The architecture using its preprocessing techniques not only takes into account the enormous variations observed in SMS data but also makes it efficient for its on-device (mobile phone) functionalities in terms of inference timing and size.
The Hyperspectral image (HSI) classification is a standard remote sensing task, in which each image pixel is given a label indicating the physical land-cover on the earth's surface. The achievements of image semantic segmentation and deep learning approaches on ordinary images have accelerated the research on hyperspectral image classification. Moreover, the utilization of both the spectral and spatial cues in hyperspectral images has shown improved classification accuracy in hyperspectral image classification. The use of only 3D Convolutional Neural Networks (3D-CNN) to extract both spatial and spectral cues from Hyperspectral images results in an explosion of parameters hence high computational cost. We propose network architecture called the MixedSN that utilizes the 3D convolutions to modeling spectral-spatial information in the early layers of the architecture and the 2D convolutions at the top layers which majorly deal with semantic abstraction. We constrain our architecture to ResNeXt block because of their performance and simplicity. Our model drastically reduced the number of parameters and achieved comparable classification performance with state-of-the-art methods on Indian Pine (IP) scene dataset, Pavia University scene (PU) dataset, Salinas (SA) Scene dataset, and Botswana (BW) dataset.
While classic studies proved that wide networks allow universal approximation, recent research and successes of deep learning demonstrate the power of deep networks. Based on a symmetric consideration, we investigate if the design of artificial neural networks should have a directional preference, and what the mechanism of interaction is between the width and depth of a network. Inspired by the De Morgan law, we address this fundamental question by establishing a quasi-equivalence between the width and depth of ReLU networks in two aspects. First, we formulate two transforms for mapping an arbitrary ReLU network to a wide network and a deep network respectively for either regression or classification so that the essentially same capability of the original network can be implemented. Then, we replace the mainstream artificial neuron type with a quadratic counterpart, and utilize the factorization and continued fraction representations of the same polynomial function to construct a wide network and a deep network, respectively. Based on our findings, a deep network has a wide equivalent, and vice versa, subject to an arbitrarily small error.