Northwestern Polytechnical University

Unsupervised learning with generative adversarial networks (GANs) has proven hugely successful. Regular GANs hypothesize the discriminator as a classifier with the sigmoid cross entropy loss function. However, we found that this loss function may lead to the vanishing gradients problem during the learning process. To overcome such a problem, we propose in this paper the Least Squares Generative Adversarial Networks (LSGANs) which adopt the least squares loss function for the discriminator. We show that minimizing the objective function of LSGAN yields minimizing the Pearson X2 divergence. There are two benefits of LSGANs over regular GANs. First, LSGANs are able to generate higher quality images than regular GANs. Second, LSGANs perform more stable during the learning process. We evaluate LSGANs on LSUN and CIFAR-10 datasets and the experimental results show that the images generated by LSGANs are of better quality than the ones generated by regular GANs. We also conduct two comparison experiments between LSGANs and regular GANs to illustrate the stability of LSGANs.

Remote sensing image scene classification plays an important role in a wide range of applications and hence has been receiving remarkable attention. During the past years, significant efforts have been made to develop various data sets or present a variety of approaches for scene classification from remote sensing images. However, a systematic review of the literature concerning data sets and methods for scene classification is still lacking. In addition, almost all existing data sets have a number of limitations, including the small scale of scene classes and the image numbers, the lack of image variations and diversity, and the saturation of accuracy. These limitations severely limit the development of new approaches especially deep learning-based methods. This paper first provides a comprehensive review of the recent progress. Then, we propose a large-scale data set, termed “NWPU-RESISC45,” which is a publicly available benchmark for REmote Sensing Image Scene Classification (RESISC), created by Northwestern Polytechnical University (NWPU). This data set contains 31 500 images, covering 45 scene classes with 700 images in each class. The proposed NWPU-RESISC45 1) is large-scale on the scene classes and the total image number; 2) holds big variations in translation, spatial resolution, viewpoint, object pose, illumination, background, and occlusion; and 3) has high within-class diversity and between-class similarity. The creation of this data set will enable the community to develop and evaluate various data-driven algorithms. Finally, several representative methods are evaluated using the proposed data set, and the results are reported as a useful baseline for future research.

Object detection in very high resolution optical remote sensing images is a fundamental problem faced for remote sensing image analysis. Due to the advances of powerful feature representations, machine-learning-based object detection is receiving increasing attention. Although numerous feature representations exist, most of them are handcrafted or shallow-learning-based features. As the object detection task becomes more challenging, their description capability becomes limited or even impoverished. More recently, deep learning algorithms, especially convolutional neural networks (CNNs), have shown their much stronger feature representation power in computer vision. Despite the progress made in nature scene images, it is problematic to directly use the CNN feature for object detection in optical remote sensing images because it is difficult to effectively deal with the problem of object rotation variations. To address this problem, this paper proposes a novel and effective approach to learn a rotation-invariant CNN (RICNN) model for advancing the performance of object detection, which is achieved by introducing and learning a new rotation-invariant layer on the basis of the existing CNN architectures. However, different from the training of traditional CNN models that only optimizes the multinomial logistic regression objective, our RICNN model is trained by optimizing a new objective function via imposing a regularization constraint, which explicitly enforces the feature representations of the training samples before and after rotating to be mapped close to each other, hence achieving rotation invariance. To facilitate training, we first train the rotation-invariant layer and then domain-specifically fine-tune the whole RICNN network to further boost the performance. Comprehensive evaluations on a publicly available ten-class object detection data set demonstrate the effectiveness of the proposed method.

Laser-metal additive manufacturing capabilities have advanced from single-material printing to multimaterial/multifunctional design and manufacturing. Material-structure-performance integrated additive manufacturing (MSPI-AM) represents a path toward the integral manufacturing of end-use components with innovative structures and multimaterial layouts to meet the increasing demand from industries such as aviation, aerospace, automobile manufacturing, and energy production. We highlight two methodological ideas for MSPI-AM-"the right materials printed in the right positions" and "unique structures printed for unique functions"-to realize major improvements in performance and function. We establish how cross-scale mechanisms to coordinate nano/microscale material development, mesoscale process monitoring, and macroscale structure and performance control can be used proactively to achieve high performance with multifunctionality. MSPI-AM exemplifies the revolution of design and manufacturing strategies for AM and its technological enhancement and sustainable development.

The unavailability of clean drinking water is one of the significant health issues in modern times. Industrial dyes are one of the dominant chemicals that make water unfit for drinking. Among these dyes, methylene blue (MB) is toxic, carcinogenic, and non-biodegradable and can cause a severe threat to human health and environmental safety. It is usually released in natural water sources, which becomes a health threat to human beings and living organisms. Hence, there is a need to develop an environmentally friendly, efficient technology for removing MB from wastewater. Photodegradation is an advanced oxidation process widely used for MB removal. It has the advantages of complete mineralization of dye into simple and nontoxic species with the potential to decrease the processing cost. This review provides a tutorial basis for the readers working in the dye degradation research area. We not only covered the basic principles of the process but also provided a wide range of previously published work on advanced photocatalytic systems (single-component and multi-component photocatalysts). Our study has focused on critical parameters that can affect the photodegradation rate of MB, such as photocatalyst type and loading, irradiation reaction time, pH of reaction media, initial concentration of dye, radical scavengers and oxidising agents. The photodegradation mechanism, reaction pathways, intermediate products, and final products of MB are also summarized. An overview of the future perspectives to utilize MB at an industrial scale is also provided. This paper identifies strategies for the development of effective MB photodegradation systems.

Metal-organic frameworks (MOFs) have been widely recognized as one of the most fascinating classes of materials from science and engineering perspectives, benefiting from their high porosity and well-defined and tailored structures and components at the atomic level. Although their intrinsic micropores endow size-selective capability and high surface area, etc., the narrow pores limit their applications toward diffusion-control and large-size species involved processes. In recent years, the construction of hierarchically porous MOFs (HP-MOFs), MOF-based hierarchically porous composites, and MOF-based hierarchically porous derivatives has captured widespread interest to extend the applications of conventional MOF-based materials. In this Review, the recent advances in the design, synthesis, and functional applications of MOF-based hierarchically porous materials are summarized. Their structural characters toward various applications, including catalysis, gas storage and separation, air filtration, sewage treatment, sensing and energy storage, have been demonstrated with typical reports. The comparison of HP-MOFs with traditional porous materials (e.g., zeolite, porous silica, carbons, metal oxides, and polymers), subsisting challenges, as well as future directions in this research field, are also indicated.

Electromagnetic (EM) wave absorbing materials play an increasingly important role in modern society for their multi-functional in military stealth and incoming 5G smart era. Dielectric loss EM wave absorbers and underlying loss mechanism investigation are of great significance to unveil EM wave attenuation behaviors of materials and guide novel dielectric loss materials design. However, current researches focus more on materials synthesis rather than in-depth mechanism study. Herein, comprehensive views toward dielectric loss mechanisms including interfacial polarization, dipolar polarization, conductive loss, and defect-induced polarization are provided. Particularly, some misunderstandings and ambiguous concepts for each mechanism are highlighted. Besides, in-depth dielectric loss study and novel dielectric loss mechanisms are emphasized. Moreover, new dielectric loss mechanism regulation strategies instead of regular components compositing are summarized to provide inspiring thoughts toward simple and effective EM wave attenuation behavior modulation.

Recent research has shown that using spectral–spatial information can considerably improve the performance of hyperspectral image (HSI) classification. HSI data is typically presented in the format of 3D cubes. Thus, 3D spatial filtering naturally offers a simple and effective method for simultaneously extracting the spectral–spatial features within such images. In this paper, a 3D convolutional neural network (3D-CNN) framework is proposed for accurate HSI classification. The proposed method views the HSI cube data altogether without relying on any preprocessing or post-processing, extracting the deep spectral–spatial-combined features effectively. In addition, it requires fewer parameters than other deep learning-based methods. Thus, the model is lighter, less likely to over-fit, and easier to train. For comparison and validation, we test the proposed method along with three other deep learning-based HSI classification methods—namely, stacked autoencoder (SAE), deep brief network (DBN), and 2D-CNN-based methods—on three real-world HSI datasets captured by different sensors. Experimental results demonstrate that our 3D-CNN-based method outperforms these state-of-the-art methods and sets a new record.

Remote sensing image scene classification is an active and challenging task driven by many applications. More recently, with the advances of deep learning models especially convolutional neural networks (CNNs), the performance of remote sensing image scene classification has been significantly improved due to the powerful feature representations learnt through CNNs. Although great success has been obtained so far, the problems of within-class diversity and between-class similarity are still two big challenges. To address these problems, in this paper, we propose a simple but effective method to learn discriminative CNNs (D-CNNs) to boost the performance of remote sensing image scene classification. Different from the traditional CNN models that minimize only the cross entropy loss, our proposed D-CNN models are trained by optimizing a new discriminative objective function. To this end, apart from minimizing the classification error, we also explicitly impose a metric learning regularization term on the CNN features. The metric learning regularization enforces the D-CNN models to be more discriminative so that, in the new D-CNN feature spaces, the images from the same scene class are mapped closely to each other and the images of different classes are mapped as farther apart as possible. In the experiments, we comprehensively evaluate the proposed method on three publicly available benchmark data sets using three off-the-shelf CNN models. Experimental results demonstrate that our proposed D-CNN methods outperform the existing baseline methods and achieve state-of-the-art results on all three data sets.

Research on sensor-based activity recognition has, recently, made significant progress and is attracting growing attention in a number of disciplines and application domains. However, there is a lack of high-level overview on this topic that can inform related communities of the research state of the art. In this paper, we present a comprehensive survey to examine the development and current status of various aspects of sensor-based activity recognition. We first discuss the general rationale and distinctions of vision-based and sensor-based activity recognition. Then, we review the major approaches and methods associated with sensor-based activity monitoring, modeling, and recognition from which strengths and weaknesses of those approaches are highlighted. We make a primary distinction in this paper between data-driven and knowledge-driven approaches, and use this distinction to structure our survey. We also discuss some promising directions for future research.

This paper proposes a double-sided LCC compensation network and its tuning method for wireless power transfer (WPT). With the proposed topology and its tuning method, the resonant frequency is irrelevant with the coupling coefficient between the two coils and is also independent of the load condition, which means that the system can work at a constant switching frequency. Analysis in frequency domain is given to show the characteristics of the proposed method. We also propose a method to tune the network to realize zero voltage switching (ZVS) for the Primary-side switches. Simulation and experimental results verified analysis and validity of the proposed compensation network and the tuning method. A wireless charging system with output power of up to 7.7 kW for electric vehicles was built, and 96% efficiency from dc power source to battery load is achieved.

We have developed a biocompatible antibacterial system based on polyethylene glycol functionalized molybdenum disulfide nanoflowers (PEG-MoS2 NFs). The PEG-MoS2 NFs have high near-infrared (NIR) absorption and peroxidase-like activity, which can efficiently catalyze decomposition of low concentration of H2O2 to generate hydroxyl radicals (·OH). The conversion of H2O2 into ·OH can avoid the toxicity of high concentration of H2O2 and the ·OH has higher antibacterial activity, making resistant bacteria more vulnerable and wounds more easily cured. The PEG-MoS2 NFs combine the catalysis with NIR photothermal effect, providing a rapid and effective killing outcome in vitro for Gram-negative ampicillin resistant Escherichia coli (Ampr E. coli) and Gram-positive endospore-forming Bacillus subtilis (B. subtilis) as compared to catalytic treatment or photothermal therapy (PTT) alone. Wound healing results indicate that the synergy antibacterial system could be conveniently used for wound disinfection in vivo. Interestingly, glutathione (GSH) oxidation can be accelerated due to the 808 nm irradiation induced hyperthermia at the presence of PEG-MoS2 NFs proved by X-ray near-edge absorption spectra and X-ray spectroscopy. The accelerated GSH oxidation can result in bacterial death more easily. A mechanism based on ·OH-enhanced PTT is proposed to explain the antibacterial process.

Electromagnetic (EM) absorbing and shielding composites with tunable absorbing behaviors based on Ti3C2 MXenes are fabricated via HF etching and annealing treatment. Localized sandwich structure without sacrificing the original layered morphology is realized, which is responsible for the enhancement of EM absorbing capability in the X-band. The composite with 50 wt % annealed MXenes exhibits a minimum reflection loss of -48.4 dB at 11.6 GHz, because of the formation of TiO2 nanocrystals and amorphous carbon. Moreover, superior shielding effectiveness with high absorption effectiveness is achieved. The total and absorbing shielding effectiveness of Ti3C2 MXenes in a wax matrix with a thickness of only 1 mm reach values of 76.1 and 67.3 dB, while those of annealed Ti3C2 MXenes/wax composites are 32 and 24.2 dB, respectively. Considering the promising performance of Ti3C2 MXenes with the modified surface, this work is expected to open the door for the expanded applications of MXenes family in EM absorbing and shielding fields.

Remote sensing image scene classification, which aims at labeling remote sensing images with a set of semantic categories based on their contents, has broad applications in a range of fields. Propelled by the powerful feature learning capabilities of deep neural networks, remote sensing image scene classification driven by deep learning has drawn remarkable attention and achieved significant breakthroughs. However, to the best of our knowledge, a comprehensive review of recent achievements regarding deep learning for scene classification of remote sensing images is still lacking. Considering the rapid evolution of this field, this article provides a systematic survey of deep learning methods for remote sensing image scene classification by covering more than 160 papers. To be specific, we discuss the main challenges of remote sensing image scene classification and survey: first, autoencoder-based remote sensing image scene classification methods; second, convolutional neural network-based remote sensing image scene classification methods; and third, generative adversarial network-based remote sensing image scene classification methods. In addition, we introduce the benchmarks used for remote sensing image scene classification and summarize the performance of more than two dozen of representative algorithms on three commonly used benchmark datasets. Finally, we discuss the promising opportunities for further research.

We report here a simple, high-yield yet low-cost approach to design single-layer MoS2 nanosheets with controllable size via an improved oleum treatment exfoliation process. By decorating MoS2 nanosheets with chitosan, these functionalized MoS2 nanosheets have been developed as a chemotherapeutic drug nanocarrier for near-infrared (NIR) photothermal-triggered drug delivery, facilitating the combination of chemotherapy and photothermal therapy into one system for cancer therapy. Loaded doxorubicin could be controllably released upon the photothermal effect induced by 808 nm NIR laser irradiation. In vitro and in vivo tumor ablation studies demonstrate a better synergistic therapeutic effect of the combined treatment, compared with either chemotherapy or photothermal therapy alone. Finally, MoS2 nanosheets can also be used as a promising contrast agent in X-ray computed tomography imaging due to the obvious X-ray absorption ability of Mo. As a result, the high-throughput oleum treatment exfoliation process could be extended for fabricating other 2D nanomaterials, and the NIR-triggered drug release strategy was encouraging for simultaneous imaging-guided cancer theranostic application.

Atomically thin boron nitride (BN) nanosheets are important two-dimensional nanomaterials with many unique properties distinct from those of graphene, but investigation into their mechanical properties remains incomplete. Here we report that high-quality single-crystalline mono- and few-layer BN nanosheets are one of the strongest electrically insulating materials. More intriguingly, few-layer BN shows mechanical behaviours quite different from those of few-layer graphene under indentation. In striking contrast to graphene, whose strength decreases by more than 30% when the number of layers increases from 1 to 8, the mechanical strength of BN nanosheets is not sensitive to increasing thickness. We attribute this difference to the distinct interlayer interactions and hence sliding tendencies in these two materials under indentation. The significantly better interlayer integrity of BN nanosheets makes them a more attractive candidate than graphene for several applications, for example, as mechanical reinforcements.

Despite being one of the oldest materials described in the chemical literature, graphitic carbon nitride (g-C3N4) has just recently experienced a renaissance as a highly active photocatalyst, and the metal-free polymer was shown to be able to generate hydrogen under visible light. The semiconductor nature of g-C3N4 has triggered tremendous endeavors on its structural manipulation for enhanced photo(electro)chemical performance, aiming at an affordable clean energy future. While pursuing the stem of g-C3N4 related catalysis (photocatalysis, electrocatalysis and photoelectrocatalysis), a number of emerging intrinsic properties of g-C3N4 are certainly interesting, but less well covered, and we believe that these novel applications outside of conventional catalysis can be favorably exploited as well. Thanks to the general efforts devoted to the exploration and enrichment of g-C3N4 based chemistry, the boundaries of this area have been possibly pushed far beyond what people could imagine in the beginning. This review strives to cover the achievements of g-C3N4 related materials in these unconventional application fields for depicting the broader future of these metal-free and fully stable semiconductors. This review starts with the general protocols to engineer g-C3N4 micro/nanostructures for practical use, and then discusses the newly disclosed applications in sensing, bioimaging, novel solar energy exploitation including photocatalytic coenzyme regeneration, templating, and carbon nitride based devices. Finally, we attempt an outlook on possible further developments in g-C3N4 based research.

Abstract Rational manipulation of hollow structure with uniform heterojunctions is evolving as an effective approach to meet the lightweight and high‐performance microwave absorption for metal‐organic frameworks (MOFs) derived absorbers. Herein, a new and controlled synergistic protecting‐etching strategy is proposed to construct shelled ZIF‐67 rhombic dodecahedral cages using tannic acid under theoretical guidance, then hollow Co@N‐doped carbon nanocages with uniform heterojunctions and hierarchical micro‐meso‐macropores are obtained via a pyrolysis process, which addresses the shortcomings of using sacrificing templates or corrosive agents. The outer Co@N‐doped carbon shell, composed of highly dispersive core‐shell heterojunctions, possesses micro‐mesopores while the inner hollow macroporous cavity endows the absorbers with lightweight characteristics. Accordingly, the maximum reflection loss is −60.6 dB at 2.4 mm and the absorption bandwidth reaches 5.1 GHz at 1.9 mm with 10 wt% filler loading, exhibiting superior specific reflection loss compared with the vast majority of previous MOFs derived absorbers. Furthermore, this synergistic protecting‐etching strategy provides inspiration for precisely creating a hollow void inside other MOFs crystals and broadens the desirable candidates for lightweight and high‐efficient microwave absorbers.

Smart nanoparticles, which can respond to biological cues or be guided by them, are emerging as a promising drug delivery platform for precise cancer treatment. The field of oncology, nanotechnology, and biomedicine has witnessed rapid progress, leading to innovative developments in smart nanoparticles for safer and more effective cancer therapy. In this review, we will highlight recent advancements in smart nanoparticles, including polymeric nanoparticles, dendrimers, micelles, liposomes, protein nanoparticles, cell membrane nanoparticles, mesoporous silica nanoparticles, gold nanoparticles, iron oxide nanoparticles, quantum dots, carbon nanotubes, black phosphorus, MOF nanoparticles, and others. We will focus on their classification, structures, synthesis, and intelligent features. These smart nanoparticles possess the ability to respond to various external and internal stimuli, such as enzymes, pH, temperature, optics, and magnetism, making them intelligent systems. Additionally, this review will explore the latest studies on tumor targeting by functionalizing the surfaces of smart nanoparticles with tumor-specific ligands like antibodies, peptides, transferrin, and folic acid. We will also summarize different types of drug delivery options, including small molecules, peptides, proteins, nucleic acids, and even living cells, for their potential use in cancer therapy. While the potential of smart nanoparticles is promising, we will also acknowledge the challenges and clinical prospects associated with their use. Finally, we will propose a blueprint that involves the use of artificial intelligence-powered nanoparticles in cancer treatment applications. By harnessing the potential of smart nanoparticles, this review aims to usher in a new era of precise and personalized cancer therapy, providing patients with individualized treatment options.

Graphene quantum dots (GQDs) that are flat 0D nanomaterials have attracted increasing interest because of their exceptional chemicophysical properties and novel applications in energy conversion and storage, electro/photo/chemical catalysis, flexible devices, sensing, display, imaging, and theranostics. The significant advances in the recent years are summarized with comparative and balanced discussion. The differences between GQDs and other nanomaterials, including their nanocarbon cousins, are emphasized, and the unique advantages of GQDs for specific applications are highlighted. The current challenges and outlook of this growing field are also discussed.