Search

Unsupervised Domain Adaptation (UDA) aims to learn a classifier for the unlabeled target domain by leveraging knowledge from a labeled source domain with a different but related distribution. Many existing approaches typically learn a domain-invariant representation space by directly matching the marginal distributions of the two domains. However, they ignore exploring the underlying discriminative features of the target data and align the cross-domain discriminative features, which may lead to suboptimal performance. To tackle these two issues simultaneously, this paper presents a Joint Clustering and Discriminative Feature Alignment (JCDFA) approach for UDA, which is capable of naturally unifying the mining of discriminative features and the alignment of class-discriminative features into one single framework. Specifically, in order to mine the intrinsic discriminative information of the unlabeled target data, JCDFA jointly learns a shared encoding representation for two tasks: supervised classification of labeled source data, and discriminative clustering of unlabeled target data, where the classification of the source domain can guide the clustering learning of the target domain to locate the object category. We then conduct the cross-domain discriminative feature alignment by separately optimizing two new metrics: 1) an extended supervised contrastive learning, i.e., semi-supervised contrastive learning 2) an extended Maximum Mean Discrepancy (MMD), i.e., conditional MMD, explicitly minimizing the intra-class dispersion and maximizing the inter-class compactness. When these two procedures, i.e., discriminative features mining and alignment are integrated into one framework, they tend to benefit from each other to enhance the final performance from a cooperative learning perspective. Experiments are conducted on four real-world benchmarks ( e.g., Office-31, ImageCLEF-DA, Office-Home and VisDA-C). All the results demonstrate that our JCDFA can obtain remarkable margins over state-of-the-art domain adaptation methods. Comprehensive ablation studies also verify the importance of each key component of our proposed algorithm and the effectiveness of combining two learning strategies into a framework.

Fault detection is a fundamental requirement for Industrial Internet of Things (IIoT), such as the process industry. This article first reviews the recent studies focusing on applying the fault detection techniques to the IIoT networks. However, we find that numerous studies focus on the resource utilization and workload allocation. The fault detection toward IIoT facilities is still in its immature stage because the existing approaches are not accurate enough for the stringent fault detection in IIoT networks. To this end, we present a novel algorithm, named Gaussian Bernoulli restricted Boltzmann machines (GBRBMs)-based deep neural network (DNN), to transform the fault detection into a classification problem. The real trace-driven experiments show that the proposed scheme outperforms other baseline machine learning methods. We anticipate that this article can inspire blooming studies on the related topics of smart IIoT networks.

To build the mechanism of a SelfDN is becoming an emerging direction for future IoT. The detection of device status, such as fault or normal, is a very fundamental module in SelfDN. In this article, we first review recent studies devoted to applying fault detection techniques in IoT networks. Taking the challenge of processing the real-valued IoT data into account, we propose a novel fault detection architecture for SelfDN. Under this architecture, we present an algorithm, named GBRBM-based deep neural network with auto-encoder (i.e., GBRBM-DAE) to transform the fault detection problem into a classification problem. The real-world trace-driven experimental results show that the proposed algorithm outperforms other popular machine learning algorithms, including linear discriminant analysis, support vector machine, pure deep neural network, and so on. Finally, we summarize some open issues of this study. We expect that this article will inspire successive studies on the related topics of SelfDN.

With the increasing demand of indoor location-based services, such as tracking targets in a smart building, device-free localization technique has attracted great attentions because it can locate the targets without employing any attached devices. Due to the limited space and complexity of the indoor environment, there still exist challenges in terms of high localization accuracy and high efficiency for indoor localization. In this paper, for addressing such issues, we first convert the received signal strength (RSS) signals into image pixels. The localization problem is then formulated as an image classification problem. To well handle the variant RSS images, a deep convolutional neural network is then structured for classification. Finally, for validating the proposed scheme, two real testbeds are built in the indoor environments, including a living room and a corridor of an apartment. Experimental results show that the proposed scheme achieves good localization performance. For example, the localization accuracy can reach up to 100% in the scenario of living room and 97.6% in the corridor. Moreover, the proposed approach outperforms the methods of the K-nearest-neighbor and the support vector machines in both the noiseless and noisy environments.

Device-free localization (DFL), as an emerging technology that locates targets without any attached devices via wireless sensor networks, has spawned extensive applications in the Internet of Things (IoT) field. For DFL, a key problem is how to extract significant features to characterize raw signals with different patterns associated with different locations. To address this problem, in this paper, the DFL problem is formulated as an image classification problem. Moreover, we design a three-layer convolutional autoencoder (CAE) neural network to perform unsupervised feature extraction from raw signals followed by supervised fine-tuning for classification. The CAE combines the advantages of a convolutional neural network (CNN) and a deep autoencoder (AE) in the feature learning and signals reconstruction, which is expected to achieve good performance for DFL. The experimental results show that the proposed approach can achieve a high localization accuracy rate of 100% for a reasonable grid size on the raw real-world data, i.e., the collected raw data without added Gaussian noise, and is robust to noisy data with a signal-to-noise ratio greater than −5 dB. Additionally, its time cost for the classification of a single activity is 4 ms, which is fast enough for the IoT applications. The proposed approach outperforms the deep CNN and AE in terms of localization accuracy and robust ability against noise.

Device-free localization (DFL) locates targets without equipping with wireless devices or tag under the Internet-of-Things (IoT) architectures. As an emerging technology, DFL has spawned extensive applications in the IoT environment, such as intrusion detection, mobile robot localization, and location-based services. Current DFL-related machine learning (ML) algorithms still suffer from low localization accuracy and weak dependability/robustness because the group structure has not been considered in their location estimation, which leads to an undependable process. To overcome these challenges, we propose in this work a dependable block-sparse scheme by particularly considering the group structure of signals. An accurate and robust ML algorithm named block-sparse coding with the proximal operator (BSCPO) is proposed for DFL. In addition, a severe Gaussian noise is added in the original sensing signals for preserving network-related privacy as well as improving the dependability of the model. The real-world data-driven experimental results show that the proposed BSCPO achieves robust localization and signal-recovery performance even under severely noisy conditions and outperforms state-of-the-art DFL methods. For single-target localization, BSCPO retains high accuracy when the signal-to-noise ratio exceeds −10 dB. BSCPO is also able to localize accurately under most multitarget localization test cases.

Sparse coding, which aims at finding appropriate sparse representations of data with an overcomplete dictionary set, is a well-established signal processing methodology and has good efficiency in various areas. The varying sparse constraint can influence the performances of sparse coding algorithms greatly. However, commonly used sparse regularization may not be robust in high-coherence condition. In this paper, inspired from independently interpretable lasso (IILasso), which considers the coherence of sensing matrix columns in constraint to implement the strategy of selecting uncorrelated variables, we propose a new regularization by introducing $\ell _{p}$ norm $(0 into the regularization part of IILasso. The new regularization can efficiently enhance the performances in obtaining sparse and accurate coefficient. To solve the optimization problem with the new regularization, we propose to use the coordinate descent algorithm with weighted $\ell _{1}$ norm, named independently interpretable weighted lasso (IIWLasso), and the proximal operator, named independently interpretable iterative shrinkage thresholding algorithm (II-ISTA) and independently interpretable proximal operator for $\ell _{\frac {2}{3}}$ norm regularization (II2/3PO). We present synthetic data experiments and gene expression data experiments to validate the performance of our proposed algorithms. The experiment results show that all independently interpretable algorithms can perform better than their original ones in different coherence conditions. Among them, IIWLasso can obtain relatively best performance both in relative norm error and support error of synthetic data and misclassification error of tenfold cross-validating gene expression data.

Extracting valuable and discriminative features is one of the crucial issues for target recognition in synthetic aperture radar (SAR) images. In this study, a feature extraction method based on robust locality discriminant projection (RLDP) is presented for SAR target recognition. To characterise the local structural information of SAR images, the manifold learning technique called the supervised locality preserving projection is introduced to learn a linear projection, with which the SAR image can be cast into an implicit feature space. Then, the authors extend t-distributed stochastic neighbour embedding to a parametric framework for optimising the linear projection. In the resulting feature space, the intrinsic neighbour relation with a certain class can be preserved. In addition, the separation between different classes can be enhanced. Unlike most local manifold learning methods, the proposed method is robust to changes of the neighbour parameter. To further analyse the non-linear structure, a useful variant of RLDP named kernel RLDP (KRLDP) is proposed. KRLDP exploits RLDP in an implicit reproducing kernel Hilbert space, where the kernel-based non-linear projection is learned to capture the non-linear structural information. Extensive experiments on moving and stationary target automatic recognition databases demonstrate the effectiveness of the proposed methods.

In practical device-free localization (DFL) applications, for enlarging the monitoring area and improving localization accuracy, too many nodes need to be deployed, which results in a large volume of DFL data with high dimensions. This arises a key problem of seeking an accurate and efficient approach for DFL. In order to address this problem, this paper regards DFL as a problem of sparse-representation-based classification; builds a sparse model; and then proposes two sparse-coding-based algorithms. The first algorithm, sparse coding via the iterative shrinkage-thresholding algorithm (SC-ISTA), is efficient for handling high-dimensional data. And then, subspace techniques are further utilized, followed by performing sparse coding in the low-dimensional signal subspace, which leads to the second algorithm termed subspace-based SC-ISTA (SSC-ISTA). Experiments with the real-world data set are conducted for single-target and multi-target localization, and three typical machine learning algorithms, deep learning based on auto encoder, K-nearest neighbor, and orthogonal matching pursuit, are compared. Experimental results show that both SC-ISTA and SSC-ISTA can achieve high localization accuracies of 100% and are robust to noisy data when SNR is greater than 10 dB, and the time costs for sparse coding of SC-ISTA and SSC-ISTA are 2.1 × 10-3 s and 2.1 × 10-4 s respectively, which indicates that the proposed algorithms outperform the other three ones.

This paper proposes a new method for feature extraction of synthetic aperture radar (SAR) image based on parametric supervised t-stochastic neighbor embedding (PS-tSNE). To mitigate the rejection of dissimilar targets from the same class, aspect angles of targets are modeled to supervise and construct the distributional properties of the training data in the original space. Then, an explicit nonlinear mapping using kernel trick is proposed by an extension of non-parametric t-SNE supervised by the information of aspect angles. This method preserves the local structure of the targets of SAR images as well as possible and enables explicit out-of-sample extensions. Experimental results based on moving and stationary target automatic recognition (MSTAR) dataset illustrate the effective performance of the proposed method on visualization and recognition.

As an emerging technology, device-free localization (DFL), using wireless sensor network to detect targets who do not need carry any attached devices, has spawned extensive applications, e.g., intrusion detection or tracking in security safeguards. Many previous studies formulate DFL as a classification problem, but there are still several challenges in terms of accuracy, robustness, etc. In this paper, we exploit a new log-regularizer in the objective function for classification. With taking the distinctive ability of log-regularizer to measure sparsity, the proposed approach can achieve an accurate localization process with robust performance in the challenging environments. Even if the input data is severely polluted by noise with a level of SNR = −10 dB, our algorithm can still keep a high accuracy of 99.4%, which outperforms five other machine learning algorithms, e.g., deep auto-encoder, convolutional neural network, etc.

With the rapid development of information processing technology and continuously evolving technological hardware and software, the scale of spatial data has grown exponentially. This has necessitated that stricter requirements be placed on the way data is processed. The traditional single-machine centralized data processing method is gradually being replaced by the network-based cloud geographic information system (GIS) mode. However, the information processing method of the business system in the existing spatial information network environment is difficult to expand, which limits the application of the business system. This paper proposes an on-demand and extended model for the GIS data processing procedure that considers the three-way separation of algorithm development, business processes, and the operational interface, and can implement an on-demand expansion of business processes by process modeling the business and task scheduling the workflow engine. This can expand the scope of business systems and improve the efficiency of business system construction.

In this paper, a new method of site model based change detection is presented for multitemporal synthetic aperture radar (SAR) images. It first constructs a site model offline making use of a high resolution image of the fixed site, and then accurate registration is carried out between the model and the images. With the location information contained in the site model, the region of interest (ROI) can be extracted easily and robustly. Finally, the significant changes are obtained by comparing the invariant features extracted separately from the shapes of targets in ROI of two images. The experiment of change detection for two SAR images of airport demonstrated the validity of the proposed method.

Device-free localization (DFL) locates target in a wireless sensors network (WSN) without equipping with wireless devices or tags, which is an emerging technology in the fields of intrusion detection and monitoring. In order to achieve an accurate result of DFL, the conventional works adopt l 1 norm as a regularizer to take the full potential of sparsity for locating targets. Contrasting to the previous works, we exploit the l 2 , 1 norm as the regularizer and devise an efficient optimization method with a proximal operator-based scheme, which leads the proposed improved-sparse-coding algorithm with proximal operator (ISCPO). Compared with the state-of-the-art methods that adopt l 1 norm as the regularizer, the proposed algorithm can improve the joint sparsity of sparse solution. Experimental results on our real testbeds of indoor DFL show that, in scenarios of living room and corridor, the proposed approach can achieve high localization accuracies of about 100% and 90%, respectively. In addition, the proposed ISCPO algorithm outperforms the compared state-of-the-art methods and has a more robust performance in challenged environments for target localization.

Device-free localization (DFL) locates targets without being equipped with the attached devices, which is of great significance for intrusion detection or monitoring in the era of the Internet-of-Things (IoT). Aiming at solving the problems of low accuracy and low robustness in DFL approaches, in this paper, we first treat the RSS signal as an RSS-image matrix and conduct a process of eliminating the background to dig out the variation component with distinguished features. Then, we make use of these feature-rich images by formulating DFL as an image classification problem. Furthermore, a deep convolutional neural network (CNN) is designed to extract features automatically for classification. The localization performance of the proposed background elimination-based CNN (BE-CNN) scheme is validated with a real-world dataset of outdoor DFL. In addition, we also validate the robust performance of the proposal by conducting numerical experiments with different levels of noise. Experimental results demonstrate that the proposed scheme has an obvious advantage in terms of improving localization accuracy and robustness for DFL. Particularly, the BE-CNN can maintain the highest localization accuracy of 100%, even in noisy conditions when the SNR is over −5 dB. The BE-based methods can outperform all the corresponding raw data-based methods in terms of the localization accuracy. In addition, the proposed method can outperform the comparison methods, deep neural network with autoencoder, K-nearest-neighbor (KNN), support vector machines (SVM), etc., in terms of the localization accuracy and robustness.

Local simple and highly discriminative radial difference texture feature.Single Bag of Words histogram model which is time and storage saving.Significant faster than the popular sparse reconstruction approach.Proposed method: computationally very fast, no tuning of parameters.Our approach gives very robust segmentation performance based on extensive experiments. This paper develops a theoretically simple, efficient yet robust approach to supervised texture segmentation based on local radial difference features and a Bag-of-Words (BoW) model. We make the following contributions: (1) we propose an approach to optimally learn new compact single BoW histogram models from the entire training set, with the single histogram providing benefits of efficiency in both memory and computation costs; (2) we show that BoW histograms computed from local simple radial difference features can provide an accurate pixel-wise segmentation of a textured image; and (3) we investigate whether sparse reconstruction, very successful in texture classification, assists in texture segmentation, with our study demonstrating the surprising conclusion that sparse reconstruction methods actually do not improve segmentation performance.Extensive experiments on composite natural texture images demonstrate the superiority of the proposed approach over multiple state of the art texture segmentation methods. Our experimental evaluation demonstrates a significant superiority over recent popular sparse reconstruction segmentation methods in terms of computational efficiency, while outperforming or comparable in terms of segmentation accuracy.

We propose a weakly supervised neural model for Ad-hoc Cross-lingual Information Retrieval (CLIR) from low-resource languages. Low resource languages often lack relevance annotations for CLIR, and when available the training data usually has limited coverage for possible queries. In this paper, we design a model which does not require relevance annotations, instead it is trained on samples extracted from translation corpora as weak supervision. This model relies on an attention mechanism to learn spans in the foreign sentence that are relevant to the query. We report experiments on two low resource languages: Swahili and Tagalog, trained on less that 100k parallel sentences each. The proposed model achieves 19 MAP points improvement compared to using CNNs for feature extraction, 12 points improvement from machine translation-based CLIR, and up to 6 points improvement compared to probabilistic CLIR models.

As an emerging technology, device-free localization (DFL), using radio frequency (RF) sensor networks to detect targets who do not carry any attached devices, has spawned extensive applications. Many existing works formulate DFL as a classification problem, and a key problem is how to extract discriminative features to characterize the raw wireless signal. In this paper, we present an autoencoder-based deep neural network for feature extraction, moreover, multiple Gaussian Bernoulli restricted Boltzmann machines (GBRBMs) are utilized for pre-training and dimension reduction. Experiment results show that this method of GBRBM-based autoencoder (GBRBM-AE) can achieve a high accuracy and efficient performance, which outperforms the conventional autoencoder. When the dimensions of input data are reduced from 784 to 20 dims, our algorithm can maintain a high accuracy of 97.1% and is robust to noise with SNR = 5dB.

Recently, sparse representation has achieved the impressive performance on target recognition in synthetic aperture radar (SAR) image. However, the unstable and unsupervised optimization of the sparse representation may lead to undesired recognition result. In this paper, a locality-constrained and class-specific sparse representation (LCSR) framework is presented to alleviate these problems. Instead of the sparse constraint, the locality constraint is designed to utilize the local structure information of the training samples. It provides stable representation for the samples with minor variations, which is beneficial to classification. To further improve the recognition performance, the query sample is represented as a linear combination of class-specific galleries based on the supervision of class information. The inference is reached corresponding to the class with the minimum reconstruction error. The experimental results demonstrate the effectiveness and robustness of the proposed method.

In this paper, synthetic aperture radar (SAR) image target recognition via joint manifold regularized low-rank matrix approximation (JMLMA) is presented. To capture the low-dimensional representation of SAR images, the low-rank matrix approxi mation framework is employed. However, in the actual application, targets are classified in the presence of variation in configuration and articulation, thus the underling manifold structure information may be missing in the learning process. To solve the problem, a joint manifold regularization term formed with different manifold models is proposed and incorporated into the low-rank matrix approximation framework. Hence, the pro posed method can not only obtain the low-dimension representation of SAR images, but also capture the intrinsic manifold structure in samples. We conduct experiments on pub licly available MSTAR database to verify the the effectiveness of the proposed method.

With the increasing interest in large-scale, high-resolution and real-time geographic information system (GIS) applications and spatial big data processing, traditional GIS is not efficient enough to handle the required loads due to limited computational capabilities.Various attempts have been made to adopt high performance computation techniques from different applications, such as designs of advanced architectures, strategies of data partition and direct parallelization method of spatial analysis algorithm, to address such challenges. This paper surveys the current state of parallel GIS with respect to parallel GIS architectures, parallel processing strategies, and relevant topics. We present the general evolution of the GIS architecture which includes main two parallel GIS architectures based on high performance computing cluster and Hadoop cluster. Then we summarize the current spatial data partition strategies, key methods to realize parallel GIS in the view of data decomposition and progress of the special parallel GIS algorithms. We use the parallel processing of GRASS as a case study. We also identify key problems and future potential research directions of parallel GIS. Refereed/Peer-reviewed

Shape is considered to be one of the most promising tools to represent and recognize an object. In this paper, an effective and rigorous shape descriptor is developed for shape recognition under affine transformations. To reflect the gray information of the shape, two extended centroids with different orders are adopted to form the vector which is treated as the x-axis to establish a new positively oriented orthonormal frame. Then, the descriptor representing the distribution of all points within the shape region along the y-axis is captured to achieve shape recognition. In addition to its stable affine invariance and robustness to noise, the descriptor has low computational complexity. The proposed descriptor has been implemented on a famous database and gives encouraging results while it is compared to other two state-of-the-art algorithms including GC and EC-ARC.

This paper proposes a classification method based on principal component reconstruction (PCR) for target recognition in synthetic aperture radar (SAR) image. To characterize the SAR image and alleviate the influence of different intensity of the same targets on target recognition, the SAR image is mapped into the principal component space by the principal component analysis with zero mean. In the principal component space, the query sample is reconstructed by a combination of the nearest training samples from the specific class, respectively. Then, the identity is decided by evaluating the minimum reconstruction error with respect to each class. Specifically, the reconstruction coefficients are constrained to be positive, which is more reasonable to the certain application. Extensive experiments carried out on the moving and stationary target acquisition and recognition (MSTAR) dataset illustrate the effective performance of the proposed method on ten-class targets recognition and configuration variation.