PyTorchKR Papers

최신 논문

총 11,809편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:1607.05440

Collaborative Layer-wise Discriminative Learning in Deep Neural Networks

Intermediate features at different layers of a deep neural network are known to be discriminative for visual patterns of different complexities. However, most existing works ignore such cross-layer heterogeneities when classifying samples of different complexities. For example, if a training sample has already been correctly classified at a specific layer with high confidence, we argue that it is unnecessary to enforce rest layers to classify this sample correctly and a better strategy is to encourage those layers to focus on other samples. In this paper, we propose a layer-wise discriminative learning method to enhance the discriminative capability of a deep network by allowing its layers to work collaboratively for classification. Towards this target, we introduce multiple classifiers on top of multiple layers. Each classifier not only tries to correctly classify the features from its input layer, but also coordinates with other classifiers to jointly maximize the final classification performance. Guided by the other companion classifiers, each classifier learns to concentrate on certain training examples and boosts the overall performance. Allowing for end-to-end training, our method can be conveniently embedded into state-of-the-art deep networks. Experiments with multiple popular deep networks, including Network in Network, GoogLeNet and VGGNet, on scale-various object classification benchmarks, including CIFAR100, MNIST and ImageNet, and scene classification benchmarks, including MIT67, SUN397 and Places205, demonstrate the effectiveness of our method. In addition, we also analyze the relationship between the proposed method and classical conditional random fields models.

arXiv:1604.03073

Reservoir computing for spatiotemporal signal classification without trained output weights

Reservoir computing is a recently introduced machine learning paradigm that has been shown to be well-suited for the processing of spatiotemporal data. Rather than training the network node connections and weights via backpropagation in traditional recurrent neural networks, reservoirs instead have fixed connections and weights among the `hidden layer' nodes, and traditionally only the weights to the output layer of neurons are trained using linear regression. We claim that for signal classification tasks one may forgo the weight training step entirely and instead use a simple supervised clustering method based upon principal components of norms of reservoir states. The proposed method is mathematically analyzed and explored through numerical experiments on real-world data. The examples demonstrate that the proposed may outperform the traditional trained output weight approach in terms of classification accuracy and sensitivity to reservoir parameters.

arXiv:1607.06525공식 코드 · Papers with Code

CGMOS: Certainty Guided Minority OverSampling

Handling imbalanced datasets is a challenging problem that if not treated correctly results in reduced classification performance. Imbalanced datasets are commonly handled using minority oversampling, whereas the SMOTE algorithm is a successful oversampling algorithm with numerous extensions. SMOTE extensions do not have a theoretical guarantee during training to work better than SMOTE and in many instances their performance is data dependent. In this paper we propose a novel extension to the SMOTE algorithm with a theoretical guarantee for improved classification performance. The proposed approach considers the classification performance of both the majority and minority classes. In the proposed approach CGMOS (Certainty Guided Minority OverSampling) new data points are added by considering certainty changes in the dataset. The paper provides a proof that the proposed algorithm is guaranteed to work better than SMOTE for training data. Further experimental results on 30 real-world datasets show that CGMOS works better than existing algorithms when using 6 different classifiers.

arXiv:1606.07908

Label Tree Embeddings for Acoustic Scene Classification

We present in this paper an efficient approach for acoustic scene classification by exploring the structure of class labels. Given a set of class labels, a category taxonomy is automatically learned by collectively optimizing a clustering of the labels into multiple meta-classes in a tree structure. An acoustic scene instance is then embedded into a low-dimensional feature representation which consists of the likelihoods that it belongs to the meta-classes. We demonstrate state-of-the-art results on two different datasets for the acoustic scene classification task, including the DCASE 2013 and LITIS Rouen datasets.

arXiv:1606.06083

Product Classification in E-Commerce using Distributional Semantics

Product classification is the task of automatically predicting a taxonomy path for a product in a predefined taxonomy hierarchy given a textual product description or title. For efficient product classification we require a suitable representation for a document (the textual description of a product) feature vector and efficient and fast algorithms for prediction. To address the above challenges, we propose a new distributional semantics representation for document vector formation. We also develop a new two-level ensemble approach utilizing (with respect to the taxonomy tree) a path-wise, node-wise and depth-wise classifiers for error reduction in the final product classification. Our experiments show the effectiveness of the distributional representation and the ensemble approach on data sets from a leading e-commerce platform and achieve better results on various evaluation metrics compared to earlier approaches.

arXiv:1605.05414

On the Evaluation of Dialogue Systems with Next Utterance Classification

An open challenge in constructing dialogue systems is developing methods for automatically learning dialogue strategies from large amounts of unlabelled data. Recent work has proposed Next-Utterance-Classification (NUC) as a surrogate task for building dialogue systems from text data. In this paper we investigate the performance of humans on this task to validate the relevance of NUC as a method of evaluation. Our results show three main findings: (1) humans are able to correctly classify responses at a rate much better than chance, thus confirming that the task is feasible, (2) human performance levels vary across task domains (we consider 3 datasets) and expertise levels (novice vs experts), thus showing that a range of performance is possible on this type of task, (3) automated dialogue systems built using state-of-the-art machine learning methods have similar performance to the human novices, but worse than the experts, thus confirming the utility of this class of tasks for driving further research in automated dialogue systems.

arXiv:1507.04126

Untangling AdaBoost-based Cost-Sensitive Classification. Part II: Empirical Analysis

A lot of approaches, each following a different strategy, have been proposed in the literature to provide AdaBoost with cost-sensitive properties. In the first part of this series of two papers, we have presented these algorithms in a homogeneous notational framework, proposed a clustering scheme for them and performed a thorough theoretical analysis of those approaches with a fully theoretical foundation. The present paper, in order to complete our analysis, is focused on the empirical study of all the algorithms previously presented over a wide range of heterogeneous classification problems. The results of our experiments, confirming the theoretical conclusions, seem to reveal that the simplest approach, just based on cost-sensitive weight initialization, is the one showing the best and soundest results, despite having been recurrently overlooked in the literature.

arXiv:1507.04125

Untangling AdaBoost-based Cost-Sensitive Classification. Part I: Theoretical Perspective

Boosting algorithms have been widely used to tackle a plethora of problems. In the last few years, a lot of approaches have been proposed to provide standard AdaBoost with cost-sensitive capabilities, each with a different focus. However, for the researcher, these algorithms shape a tangled set with diffuse differences and properties, lacking a unifying analysis to jointly compare, classify, evaluate and discuss those approaches on a common basis. In this series of two papers we aim to revisit the various proposals, both from theoretical (Part I) and practical (Part II) perspectives, in order to analyze their specific properties and behavior, with the final goal of identifying the algorithm providing the best and soundest results.

arXiv:1607.08635

A 58.6mW Real-Time Programmable Object Detector with Multi-Scale Multi-Object Support Using Deformable Parts Model on 1920x1080 Video at 30fps

This paper presents a programmable, energy-efficient and real-time object detection accelerator using deformable parts models (DPM), with 2x higher accuracy than traditional rigid body models. With 8 deformable parts detection, three methods are used to address the high computational complexity: classification pruning for 33x fewer parts classification, vector quantization for 15x memory size reduction, and feature basis projection for 2x reduction of the cost of each classification. The chip is implemented in 65nm CMOS technology, and can process HD (1920x1080) images at 30fps without any off-chip storage while consuming only 58.6mW (0.94nJ/pixel, 1168 GOPS/W). The chip has two classification engines to simultaneously detect two different classes of objects. With a tested high throughput of 60fps, the classification engines can be time multiplexed to detect even more than two object classes. It is energy scalable by changing the pruning factor or disabling the parts classification.

arXiv:1607.08400

Randomised Algorithm for Feature Selection and Classification

We here introduce a novel classification approach adopted from the nonlinear model identification framework, which jointly addresses the feature selection and classifier design tasks. The classifier is constructed as a polynomial expansion of the original attributes and a model structure selection process is applied to find the relevant terms of the model. The selection method progressively refines a probability distribution defined on the model structure space, by extracting sample models from the current distribution and using the aggregate information obtained from the evaluation of the population of models to reinforce the probability of extracting the most important terms. To reduce the initial search space, distance correlation filtering can be applied as a preprocessing technique. The proposed method is evaluated and compared to other well-known feature selection and classification methods on standard benchmark classification problems. The results show the effectiveness of the proposed method with respect to competitor methods both in terms of classification accuracy and model complexity. The obtained models have a simple structure, easily amenable to interpretation and analysis.

arXiv:1607.08149

N-opcode Analysis for Android Malware Classification and Categorization

Malware detection is a growing problem particularly on the Android mobile platform due to its increasing popularity and accessibility to numerous third party app markets. This has also been made worse by the increasingly sophisticated detection avoidance techniques employed by emerging malware families. This calls for more effective techniques for detection and classification of Android malware. Hence, in this paper we present an n-opcode analysis based approach that utilizes machine learning to classify and categorize Android malware. This approach enables automated feature discovery that eliminates the need for applying expert or domain knowledge to define the needed features. Our experiments on 2520 samples that were performed using up to 10-gram opcode features showed that an f-measure of 98% is achievable using this approach.

arXiv:1607.07987

A Multiple Kernel Learning Approach for Human Behavioral Task Classification using STN-LFP Signal

Deep Brain Stimulation (DBS) has gained increasing attention as an effective method to mitigate Parkinsons disease (PD) disorders. Existing DBS systems are open-loop such that the system parameters are not adjusted automatically based on patients behavior. Classification of human behavior is an important step in the design of the next generation of DBS systems that are closed-loop. This paper presents a classification approach to recognize such behavioral tasks using the subthalamic nucleus (STN) Local Field Potential (LFP) signals. In our approach, we use the time-frequency representation (spectrogram) of the raw LFP signals recorded from left and right STNs as the feature vectors. Then these features are combined together via Support Vector Machines (SVM) with Multiple Kernel Learning (MKL) formulation. The MKL-based classification method is utilized to classify different tasks: button press, mouth movement, speech, and arm movement. Our experiments show that the lp-norm MKL significantly outperforms single kernel SVM-based classifiers in classifying behavioral tasks of five subjects even using signals acquired with a low sampling rate of 10 Hz. This leads to a lower computational cost.

arXiv:1607.06280

Explaining Classification Models Built on High-Dimensional Sparse Data

Predictive modeling applications increasingly use data representing people's behavior, opinions, and interactions. Fine-grained behavior data often has different structure from traditional data, being very high-dimensional and sparse. Models built from these data are quite difficult to interpret, since they contain many thousands or even many millions of features. Listing features with large model coefficients is not sufficient, because the model coefficients do not incorporate information on feature presence, which is key when analysing sparse data. In this paper we introduce two alternatives for explaining predictive models by listing important features. We evaluate these alternatives in terms of explanation "bang for the buck,", i.e., how many examples' inferences are explained for a given number of features listed. The bottom line: (i) The proposed alternatives have double the bang-for-the-buck as compared to just listing the high-coefficient features, and (ii) interestingly, although they come from different sources and motivations, the two new alternatives provide strikingly similar rankings of important features.

arXiv:1610.05174

Spatio-temporal Co-Occurrence Characterizations for Human Action Classification

The human action classification task is a widely researched topic and is still an open problem. Many state-of-the-arts approaches involve the usage of bag-of-video-words with spatio-temporal local features to construct characterizations for human actions. In order to improve beyond this standard approach, we investigate the usage of co-occurrences between local features. We propose the usage of co-occurrences information to characterize human actions. A trade-off factor is used to define an optimal trade-off between vocabulary size and classification rate. Next, a spatio-temporal co-occurrence technique is applied to extract co-occurrence information between labeled local features. Novel characterizations for human actions are then constructed. These include a vector quantized correlogram-elements vector, a highly discriminative PCA (Principal Components Analysis) co-occurrence vector and a Haralick texture vector. Multi-channel kernel SVM (support vector machine) is utilized for classification. For evaluation, the well known KTH as well as the challenging UCF-Sports action datasets are used. We obtained state-of-the-arts classification performance. We also demonstrated that we are able to fully utilize co-occurrence information, and improve the standard bag-of-video-words approach.

arXiv:1608.00501

Supervised Classification of RADARSAT-2 Polarimetric Data for Different Land Features

The pixel percentage belonging to the user defined area that are assigned to cluster in a confusion matrix for RADARSAT-2 over Vancouver area has been analysed for classification. In this study, supervised Wishart and Support Vector Machine (SVM) classifiers over RADARSAT-2 (RS2) fine quadpol mode Single Look Complex (SLC) product data is computed and compared. In comparison with conventional single channel or dual channel polarization, RADARSAT-2 is fully polarimetric, making it to offer better land feature contrast for classification operation.

arXiv:1608.00182

Deep FisherNet for Object Classification

Despite the great success of convolutional neural networks (CNN) for the image classification task on datasets like Cifar and ImageNet, CNN's representation power is still somewhat limited in dealing with object images that have large variation in size and clutter, where Fisher Vector (FV) has shown to be an effective encoding strategy. FV encodes an image by aggregating local descriptors with a universal generative Gaussian Mixture Model (GMM). FV however has limited learning capability and its parameters are mostly fixed after constructing the codebook. To combine together the best of the two worlds, we propose in this paper a neural network structure with FV layer being part of an end-to-end trainable system that is differentiable; we name our network FisherNet that is learnable using backpropagation. Our proposed FisherNet combines convolutional neural network training and Fisher Vector encoding in a single end-to-end structure. We observe a clear advantage of FisherNet over plain CNN and standard FV in terms of both classification accuracy and computational efficiency on the challenging PASCAL VOC object classification task.

arXiv:1607.08878공식 코드 · Papers with Code

Identifying and Harnessing the Building Blocks of Machine Learning Pipelines for Sensible Initialization of a Data Science Automation Tool

As data science continues to grow in popularity, there will be an increasing need to make data science tools more scalable, flexible, and accessible. In particular, automated machine learning (AutoML) systems seek to automate the process of designing and optimizing machine learning pipelines. In this chapter, we present a genetic programming-based AutoML system called TPOT that optimizes a series of feature preprocessors and machine learning models with the goal of maximizing classification accuracy on a supervised classification problem. Further, we analyze a large database of pipelines that were previously used to solve various supervised classification problems and identify 100 short series of machine learning operations that appear the most frequently, which we call the building blocks of machine learning pipelines. We harness these building blocks to initialize TPOT with promising solutions, and find that this sensible initialization method significantly improves TPOT's performance on one benchmark at no cost of significantly degrading performance on the others. Thus, sensible initialization with machine learning pipeline building blocks shows promise for GP-based AutoML systems, and should be further refined in future work.

arXiv:1607.03516

Deep Reconstruction-Classification Networks for Unsupervised Domain Adaptation

In this paper, we propose a novel unsupervised domain adaptation algorithm based on deep learning for visual object recognition. Specifically, we design a new model called Deep Reconstruction-Classification Network (DRCN), which jointly learns a shared encoding representation for two tasks: i) supervised classification of labeled source data, and ii) unsupervised reconstruction of unlabeled target data.In this way, the learnt representation not only preserves discriminability, but also encodes useful information from the target domain. Our new DRCN model can be optimized by using backpropagation similarly as the standard neural networks. We evaluate the performance of DRCN on a series of cross-domain object recognition tasks, where DRCN provides a considerable improvement (up to ~8% in accuracy) over the prior state-of-the-art algorithms. Interestingly, we also observe that the reconstruction pipeline of DRCN transforms images from the source domain into images whose appearance resembles the target dataset. This suggests that DRCN's performance is due to constructing a single composite representation that encodes information about both the structure of target images and the classification of source images. Finally, we provide a formal analysis to justify the algorithm's objective in domain adaptation context.

arXiv:1608.00853

A study of the effect of JPG compression on adversarial images

Neural network image classifiers are known to be vulnerable to adversarial images, i.e., natural images which have been modified by an adversarial perturbation specifically designed to be imperceptible to humans yet fool the classifier. Not only can adversarial images be generated easily, but these images will often be adversarial for networks trained on disjoint subsets of data or with different architectures. Adversarial images represent a potential security risk as well as a serious machine learning challenge---it is clear that vulnerable neural networks perceive images very differently from humans. Noting that virtually every image classification data set is composed of JPG images, we evaluate the effect of JPG compression on the classification of adversarial images. For Fast-Gradient-Sign perturbations of small magnitude, we found that JPG compression often reverses the drop in classification accuracy to a large extent, but not always. As the magnitude of the perturbations increases, JPG recompression alone is insufficient to reverse the effect.

arXiv:1608.00848

A New Android Malware Detection Approach Using Bayesian Classification

Mobile malware has been growing in scale and complexity as smartphone usage continues to rise. Android has surpassed other mobile platforms as the most popular whilst also witnessing a dramatic increase in malware targeting the platform. A worrying trend that is emerging is the increasing sophistication of Android malware to evade detection by traditional signature-based scanners. As such, Android app marketplaces remain at risk of hosting malicious apps that could evade detection before being downloaded by unsuspecting users. Hence, in this paper we present an effective approach to alleviate this problem based on Bayesian classification models obtained from static code analysis. The models are built from a collection of code and app characteristics that provide indicators of potential malicious activities. The models are evaluated with real malware samples in the wild and results of experiments are presented to demonstrate the effectiveness of the proposed approach.