Lu SUN Mineichi KUDO Keigo KIMURA
Multi-label classification is an appealing and challenging supervised learning problem, where multiple labels, rather than a single label, are associated with an unseen test instance. To remove possible noises in labels and features of high-dimensionality, multi-label dimension reduction has attracted more and more attentions in recent years. The existing methods usually suffer from several problems, such as ignoring label outliers and label correlations. In addition, most of them emphasize on conducting dimension reduction in an unsupervised or supervised way, therefore, unable to utilize the label information or a large amount of unlabeled data to improve the performance. In order to cope with these problems, we propose a novel method termed Robust sEmi-supervised multi-lAbel DimEnsion Reduction, shortly READER. From the viewpoint of empirical risk minimization, READER selects most discriminative features for all the labels in a semi-supervised way. Specifically, the ℓ2,1-norm induced loss function and regularization term make READER robust to the outliers in the data points. READER finds a feature subspace so as to keep originally neighbor instances close and embeds labels into a low-dimensional latent space nonlinearly. To optimize the objective function, an efficient algorithm is developed with convergence property. Extensive empirical studies on real-world datasets demonstrate the superior performance of the proposed method.
Sanay MUHAMMAD UMAR SAEED Syed MUHAMMAD ANWAR Muhammad MAJID
A study on quantification of human stress using low beta waves of electroencephalography (EEG) is presented. For the very first time the importance of low beta waves as a feature for quantification of human stress is highlighted. In this study, there were twenty-eight participants who filled the Perceived Stress Scale (PSS) questionnaire and recorded their EEG in closed eye condition by using a commercially available single channel EEG headset placed at frontal site. On the regression analysis of beta waves extracted from recorded EEG, it has been observed that low beta waves can predict PSS scores with a confidence level of 94%. Consequently, when low beta wave is used as a feature with the Naive Bayes algorithm for classification of stress level, it not only reduces the computational cost by 7 folds but also improves the accuracy to 71.4%.
Seongkyu MUN Minkyu SHIN Suwon SHON Wooil KIM David K. HAN Hanseok KO
Recent acoustic event classification research has focused on training suitable filters to represent acoustic events. However, due to limited availability of target event databases and linearity of conventional filters, there is still room for improving performance. By exploiting the non-linear modeling of deep neural networks (DNNs) and their ability to learn beyond pre-trained environments, this letter proposes a DNN-based feature extraction scheme for the classification of acoustic events. The effectiveness and robustness to noise of the proposed method are demonstrated using a database of indoor surveillance environments.
Jaekeun YUN Daehee KIM Sunshin AN
Since the sensor nodes are subject to faults due to the highly-constrained resources and hostile deployment environments, fault management in wireless sensor networks (WSNs) is essential to guarantee the proper operation of networks, especially routing. In contrast to existing fault management methods which mainly aim to be tolerant to faults without considering the fault type, we propose a novel efficient fault-aware routing method where faults are classified and dealt with accordingly. More specifically, we first identify each fault and then try to set up the new routing path according to the fault type. Our proposed method can be easily integrated with any kind of existing routing method. We show that our proposed method outperforms AODV, REAR, and GPSR, which are the representative works of single-path routing, multipath routing and location based routing, in terms of energy efficiency and data delivery ratio.
An enormous number of malware samples pose a major threat to our networked society. Antivirus software and intrusion detection systems are widely implemented on the hosts and networks as fundamental countermeasures. However, they may fail to detect evasive malware. Thus, setting a high priority for new varieties of malware is necessary to conduct in-depth analyses and take preventive measures. In this paper, we present a traffic model for malware that can classify network behaviors of malware and identify new varieties of malware. Our model comprises malware-specific features and general traffic features that are extracted from packet traces obtained from a dynamic analysis of the malware. We apply a clustering analysis to generate a classifier and evaluate our proposed model using large-scale live malware samples. The results of our experiment demonstrate the effectiveness of our model in finding new varieties of malware.
Hao GE Feng YANG Xiaoguang TU Mei XIE Zheng MA
Recently, numerous methods have been proposed to tackle the problem of fine-grained image classification. However, rare of them focus on the pre-processing step of image alignment. In this paper, we propose a new pre-processing method with the aim of reducing the variance of objects among the same class. As a result, the variance of objects between different classes will be more significant. The proposed approach consists of four procedures. The “parts” of the objects are firstly located. After that, the rotation angle and the bounding box could be obtained based on the spatial relationship of the “parts”. Finally, all the images are resized to similar sizes. The objects in the images possess the properties of translation, scale and rotation invariance after processed by the proposed method. Experiments on the CUB-200-2011 and CUB-200-2010 datasets have demonstrated that the proposed method could boost the recognition performance by serving as a pre-processing step of several popular classification algorithms.
This paper investigates the effect of noises added to hidden units of AutoEncoders linked to multilayer perceptrons. It is shown that internal representation of learned features emerges and sparsity of hidden units increases when independent Gaussian noises are added to inputs of hidden units during the deep network training. It is also shown that the weights that connect the contaminated hidden units with the next layer have smaller values and outputs of hidden units tend to be more definite (0 or 1). This is expected to improve the generalization ability of the network through this automatic structuration by adding the noises. This network structuration was confirmed by experiments for MNIST digits classification via a deep neural network model.
Tadashi MATSUO Nobutaka SHIMADA
Appearance-based generic object recognition is a challenging problem because all possible appearances of objects cannot be registered, especially as new objects are produced every day. Function of objects, however, has a comparatively small number of prototypes. Therefore, function-based classification of new objects could be a valuable tool for generic object recognition. Object functions are closely related to hand-object interactions during handling of a functional object; i.e., how the hand approaches the object, which parts of the object and contact the hand, and the shape of the hand during interaction. Hand-object interactions are helpful for modeling object functions. However, it is difficult to assign discrete labels to interactions because an object shape and grasping hand-postures intrinsically have continuous variations. To describe these interactions, we propose the interaction descriptor space which is acquired from unlabeled appearances of human hand-object interactions. By using interaction descriptors, we can numerically describe the relation between an object's appearance and its possible interaction with the hand. The model infers the quantitative state of the interaction from the object image alone. It also identifies the parts of objects designed for hand interactions such as grips and handles. We demonstrate that the proposed method can unsupervisedly generate interaction descriptors that make clusters corresponding to interaction types. And also we demonstrate that the model can infer possible hand-object interactions.
Haibo YIN Jun-an YANG Wei WANG Hui LIU
Transfer boosting, a branch of instance-based transfer learning, is a commonly adopted transfer learning method. However, currently popular transfer boosting methods focus on binary classification problems even though there are many multi-classification tasks in practice. In this paper, we developed a new algorithm called MultiTransferBoost on the basis of TransferBoost for multi-classification. MultiTransferBoost firstly separated the multi-classification problem into several orthogonal binary classification problems. During each iteration, MultiTransferBoost boosted weighted instances from different source domains while each instance's weight was assigned and updated by evaluating the difficulty of the instance being correctly classified and the “transferability” of the instance's corresponding source domain to the target. The updating process repeated until it reached the predefined training error or iteration number. The weight update factors, which were analyzed and adjusted to minimize the Hamming loss of the output coding, strengthened the connections among the sub binary problems during each iteration. Experimental results demonstrated that MultiTransferBoost had better classification performance and less computational burden than existing instance-based algorithms using the One-Against-One (OAO) strategy.
The quality of codebook is very important in visual image classification. In order to boost the classification performance, a scheme of codebook generation for scene image recognition based on parallel key SIFT analysis (PKSA) is presented in this paper. The method iteratively applies classical k-means clustering algorithm and similarity analysis to evaluate key SIFT descriptors (KSDs) from the input images, and generates the codebook by a relaxed k-means algorithm according to the set of KSDs. With the purpose of evaluating the performance of the PKSA scheme, the image feature vector is calculated by sparse code with Spatial Pyramid Matching (ScSPM) after the codebook is constructed. The PKSA-based ScSPM method is tested and compared on three public scene image datasets. The experimental results show the proposed scheme of PKSA can significantly save computational time and enhance categorization rate.
Shangqi ZHANG Haihong SHEN Chunlei HUO
Building detection from high resolution remote sensing images is challenging due to the high intraclass variability and the difficulty in describing buildings. To address the above difficulties, a novel approach is proposed based on the combination of shape-specific feature extraction and discriminative feature classification. Shape-specific feature can capture complex shapes and structures of buildings. Discriminative feature classification is effective in reflecting similarities among buildings and differences between buildings and backgrounds. Experiments demonstrate the effectiveness of the proposed approach.
Chang XU Yingguan WANG Yunlong ZHAN
This paper focus on the development of a single portable roadside magnetic sensor for vehicle classification. The magnetic sensor is a kind of anisotropic magnetic device that do not require to be embedded in the roadway-the device is placed next to the roadway and measure traffic in the immediately adjacent lane. A novel feature extraction and comparison approach is presented for vehicle classification with a single magnetic sensor, which is based on four different feature sets extracted from the detected magnetic signal. Furthermore, vehicle classification has been achieved with three common classification algorithms, including support vector machine, k-nearest neighbors and back-propagation neural network. Experimental results have demonstrated that the Peak-Peak feature set with back-propagation neural network approach performs much better than other approaches. Besides, the normalization technology has been proved it does work.
Leilei KONG Zhimao LU Zhongyuan HAN Haoliang QI
This paper addresses the issue of source retrieval in plagiarism detection. The task of source retrieval is retrieving all plagiarized sources of a suspicious document from a source document corpus whilst minimizing retrieval costs. The classification-based methods achieved the best performance in the current researches of source retrieval. This paper points out that it is more important to cast the problem as ranking and employ learning to rank methods to perform source retrieval. Specially, it employs RankBoost and Ranking SVM to obtain the candidate plagiarism source documents. Experimental results on the dataset of PAN@CLEF 2013 Source Retrieval show that the ranking based methods significantly outperforms the baseline methods based on classification. We argue that considering the source retrieval as a ranking problem is better than a classification problem.
Ordinal classification is a class of special tasks in machine learning and pattern recognition. As to ordinal classification, there is an ordinal structure among different decision values. The monotonicity constraint between features and decision should be taken into account as the fundamental assumption. However, in real-world applications, this assumption may be not true. Only some candidate features, instead of all, are monotonic with decision. So the existing feature selection algorithms which are designed for nominal classification or monotonic classification are not suitable for ordinal classification. In this paper, we propose a feature selection algorithm for ordinal classification based on considering the non-monotonic and monotonic features separately. We first introduce an assumption of hybrid monotonic classification consistency and define a feature evaluation function to calculate the relevance between the features and decision for ordinal classification. Then, we combine the reported measure and genetic algorithm (GA) to search the optimal feature subset. A collection of numerical experiments are implemented to show that the proposed approach can effectively reduce the feature size and improve the classification performance.
Chao LIANG Wenming YANG Fei ZHOU Qingmin LIAO
In this letter, we propose a novel framework to estimate the joint distribution of multiple Local Binary Patterns (LBPs). Multiple LBPs extracted from the same central pixel are first encoded using handcrafted encoding schemes to achieve rotation invariance, and the outputs are further encoded through a pre-trained Restricted Boltzmann Machine (RBM) to reduce the dimension of features. RBM has been successfully used as binary feature detectors and the binary-valued units of RBM seamlessly adapt to LBP. The proposed feature is called RBM-LBP. Experiments on the CUReT and Outex databases show that RBM-LBP is superior to conventional handcrafted encodings and more powerful in estimating the joint distribution of multiple LBPs.
Takahiro OGAWA Akihiro TAKAHASHI Miki HASEYAMA
In this paper, an insect classification method using scanning electron microphotographs is presented. Images taken by a scanning electron microscope (SEM) have a unique problem for classification in that visual features differ from each other by magnifications. Therefore, direct use of conventional methods results in inaccurate classification results. In order to successfully classify these images, the proposed method generates an optimal training dataset for constructing a classifier for each magnification. Then our method classifies images using the classifiers constructed by the optimal training dataset. In addition, several images are generally taken by an SEM with different magnifications from the same insect. Therefore, more accurate classification can be expected by integrating the results from the same insect based on Dempster-Shafer evidence theory. In this way, accurate insect classification can be realized by our method. At the end of this paper, we show experimental results to confirm the effectiveness of the proposed method.
Chenglong MA Qingwei ZHAO Jielin PAN Yonghong YAN
Short texts usually encounter the problem of data sparseness, as they do not provide sufficient term co-occurrence information. In this paper, we show how to mitigate the problem in short text classification through word embeddings. We assume that a short text document is a specific sample of one distribution in a Gaussian-Bayesian framework. Furthermore, a fast clustering algorithm is utilized to expand and enrich the context of short text in embedding space. This approach is compared with those based on the classical bag-of-words approaches and neural network based methods. Experimental results validate the effectiveness of the proposed method.
Xuyang WANG Pengyuan ZHANG Qingwei ZHAO Jielin PAN Yonghong YAN
The introduction of deep neural networks (DNNs) leads to a significant improvement of the automatic speech recognition (ASR) performance. However, the whole ASR system remains sophisticated due to the dependent on the hidden Markov model (HMM). Recently, a new end-to-end ASR framework, which utilizes recurrent neural networks (RNNs) to directly model context-independent targets with connectionist temporal classification (CTC) objective function, is proposed and achieves comparable results with the hybrid HMM/DNN system. In this paper, we investigate per-dimensional learning rate methods, ADAGRAD and ADADELTA included, to improve the recognition of the end-to-end system, based on the fact that the blank symbol used in CTC technique dominates the output and these methods give frequent features small learning rates. Experiment results show that more than 4% relative reduction of word error rate (WER) as well as 5% absolute improvement of label accuracy on the training set are achieved when using ADADELTA, and fewer epochs of training are needed.
Fine-grained visual categorization (FGVC) has drawn increasing attention as an emerging research field in recent years. In contrast to generic-domain visual recognition, FGVC is characterized by high intra-class and subtle inter-class variations. To distinguish conceptually and visually similar categories, highly discriminative visual features must be extracted. Moreover, FGVC has highly specialized and task-specific nature. It is not always easy to obtain a sufficiently large-scale training dataset. Therefore, the key to success in practical FGVC systems is to efficiently exploit discriminative features from a limited number of training examples. In this paper, we propose an efficient two-step dimensionality compression method to derive compact middle-level part-based features. To do this, we compare both space-first and feature-first convolution schemes and investigate their effectiveness. Our approach is based on simple linear algebra and analytic solutions, and is highly scalable compared with the current one-vs-one or one-vs-all approach, making it possible to quickly train middle-level features from a number of pairwise part regions. We experimentally show the effectiveness of our method using the standard Caltech-Birds and Stanford-Cars datasets.
Chao LIANG Wenming YANG Fei ZHOU Qingmin LIAO
In this letter, we propose a novel texture descriptor that takes advantage of an anisotropic neighborhood. A brand new encoding scheme called Reflection and Rotation Invariant Uniform Patterns (rriu2) is proposed to explore local structures of textures. The proposed descriptor is called Oriented Local Binary Patterns (OLBP). OLBP may be incorporated into other varieties of Local Binary Patterns (LBP) to obtain more powerful texture descriptors. Experimental results on CUReT and Outex databases show that OLBP not only significantly outperforms LBP, but also demonstrates great robustness to rotation and illuminant changes.