Search

Generative adversarial networks (GANs) directly learn from an unknown real distribution through adversarial training. However, training the generator only by the feedback of the discriminator cannot make GANs learn adaptively from the unknown complex real distribution, and for this reason the quality of generated images is unsatisfactory sometimes. To address this problem, we propose a framework for training GANs with an adaptive learning strategy from simpleness to complexity. First, we employ a pre-trained encoder and a generator to construct a simple task that looks like a real image. Second, an adaptive learning strategy is designed based on the mathematical expectation of the discriminating results of the real image and the simple task. The designed adaptive learning strategy is well compatible with various GANs architectures. Experimental results demonstrate the proposed method can improve the performance of existing GANs. [ABSTRACT FROM AUTHOR]

To investigate the application of combined audiological examination and a self-reported symptoms survey in the evaluation of hyperacusis. Patients who visited the outpatient department of Otolaryngology Head and Neck Surgery and Otological medicine, Affiliated Hospital of Southwest Medical University, from January 2019 to March 2021 were divided into a hyperacusis group and a normal control group. We measured the loudness discomfort level (LDL) and hearing threshold (HT) of the subjects and investigated their self-reported symptoms. We compared the demographic characteristics, loudness discomfort level, and hearing threshold of the two groups and analyzed the self-reported symptoms and audiological characteristics of hyperacusis. We considered 87 subjects, comprising 40 patients with hyperacusis and 47 healthy individuals. Among the hyperacusis patients, bilateral disease was predominant. Among them, 33 were females, 23 had hearing loss, and 20 had tinnitus. Patients are mainly in the 21–60 age group. Patients with hyperacusis had low discomfort thresholds at all frequencies except 500 Hz (P ≤ 0.05, mean LDL decreased by 6.14–1.37 dB HL for all frequencies). The incidences of feeling upset, pain, and anxiety or fear were 95%, 65%, and 82.5%, respectively, in patients with hyperacusis. The severity of symptoms varies between patients with hyperacusis and healthy individuals. A combination of LDL measurements and self-reported symptom surveys allows for an accurate and comprehensive assessment of hyperacusis. Trial registration: This study was retrospectively registered (TRN: ChiCTR2100047391) on June 13, 2021. [ABSTRACT FROM AUTHOR]

In this paper, we explore the verification problem of outsourcing constrained nonlinear programming (NLP) when it is required to be solved by particle swarm optimization (PSO) algorithm, i.e., making sure that the cloud runs PSO algorithm faithfully and returns an acceptable solution. An efficient verification scheme without any cryptographic tool is proposed. The proposed scheme involves approximate KKT conditions with the ε-KKT point in verifying the optimality of the result returned by PSO algorithm. Extensive experiments on PSO benchmarks and NLP test problems demonstrate that our proposed scheme is effective and efficient at verifying the cloud’s honesty. [ABSTRACT FROM AUTHOR]

In order to obtain high-quality color images, it is important to keep the hue component unchanged while emphasize the intensity or saturation component. As a public color model, Hue-Saturation Intensity (HSI) model is commonly used in image processing. A new single channel quantum color image encryption algorithm based on HSI model and quantum Fourier transform (QFT) is investigated, where the color components of the original color image are converted to HSI and the logistic map is employed to diffuse the relationship of pixels in color components. Subsequently, quantum Fourier transform is exploited to fulfill the encryption. The cipher-text is a combination of a gray image and a phase matrix. Simulations and theoretical analyses demonstrate that the proposed single channel quantum color image encryption scheme based on the HSI model and quantum Fourier transform is secure and effective. [ABSTRACT FROM AUTHOR]

Although a great number of objective methods for assessing perceptual image quality have been proposed, their performances on different distortion types are still unsatisfactory. It is a great challenge for image quality assessment (IQA) that distorted images may have similar visual perception even if their distortion types and magnitudes are totally different, because some structural changes caused by distortions are visually imperceptible. In this paper, we propose a novel full-reference IQA (FR-IQA) scheme based on multiscale fuzzy gradient similarity deviation (MFGSD), where fuzzy inference system is introduced to reduce the negative impact of imperceptible distortions, and the standard deviation of fuzzy gradient similarity is utilized to measure their quality distinction. Extensive experiments are conducted on two publicly available image databases, and compared with many existing state-of-art FR-IQA schemes, and the proposed MFGSD has a better performance on different distortion types and strengths. [ABSTRACT FROM AUTHOR]

A novel encryption algorithm for quantum images based on quantum image XOR operations is designed. The quantum image XOR operations are designed by using the hyper-chaotic sequences generated with the Chen's hyper-chaotic system to control the control-NOT operation, which is used to encode gray-level information. The initial conditions of the Chen's hyper-chaotic system are the keys, which guarantee the security of the proposed quantum image encryption algorithm. Numerical simulations and theoretical analyses demonstrate that the proposed quantum image encryption algorithm has larger key space, higher key sensitivity, stronger resistance of statistical analysis and lower computational complexity than its classical counterparts. [ABSTRACT FROM AUTHOR]

Superconductivity in copper oxide (cuprate) high-transition-temperature superconductors follows from the chemical doping of an antiferromagnetic insulating state. The consensus that the wavefunction of the superconducting carrier, the Cooper pair, has dx2−y2 symmetry has long been reached. This pairing symmetry implies the existence of nodes in the superconducting energy gap. Recently, a series of angle-resolved photoemission spectroscopy experiments have revealed that deeply underdoped cuprates exhibit a particle-hole symmetric superconducting-like energy gap at the momentum-space locations where the dx2−y2 gap nodes are expected. Here we discuss the possibility that this phenomenon is caused by a fully gapped topological superconducting state that coexists with the antiferromagnetic order. If experimentally confirmed, this result will completely change our view of how exactly the high-temperature superconductivity state evolves from the insulating antiferromagnet. [ABSTRACT FROM AUTHOR]

Classification is used to solve countless problems. Many real world computer vision problems, such as visual surveillance, contain uninteresting but common classes alongside interesting but rare classes. The rare classes are often unknown, and need to be discovered whilst training a classifier. Given a data set active learning selects the members within it to be labelled for the purpose of constructing a classifier, optimising the choice to get the best classifier for the least amount of effort. We propose an active learning method for scenarios with unknown, rare classes, where the problems of classification and rare class discovery need to be tackled jointly. By assuming a non-parametric prior on the data the goals of new class discovery and classification refinement are automatically balanced, without any tunable parameters. The ability to work with any specific classifier is maintained, so it may be used with the technique most appropriate for the problem at hand. Results are provided for a large variety of problems, demonstrating superior performance. [ABSTRACT FROM AUTHOR]

This study addresses the problem of choosing the most suitable probabilistic model selection criterion for unsupervised learning of visual context of a dynamic scene using mixture models. A rectified Bayesian Information Criterion (BICr) and a Completed Likelihood Akaike’s Information Criterion (CL-AIC) are formulated to estimate the optimal model order (complexity) for a given visual scene. Both criteria are designed to overcome poor model selection by existing popular criteria when the data sample size varies from small to large and the true mixture distribution kernel functions differ from the assumed ones. Extensive experiments on learning visual context for dynamic scene modelling are carried out to demonstrate the effectiveness of BICr and CL-AIC, compared to that of existing popular model selection criteria including BIC, AIC and Integrated Completed Likelihood (ICL). Our study suggests that for learning visual context using a mixture model, BICr is the most appropriate criterion given sparse data, while CL-AIC should be chosen given moderate or large data sample sizes. [ABSTRACT FROM AUTHOR]

Neural networks do not generalize well to unseen data with domain shifts—a longstanding problem in machine learning and AI. To overcome the problem, we propose MixStyle, a simple plug-and-play, parameter-free module that can improve domain generalization performance without the need to collect more data or increase model capacity. The design of MixStyle is simple: it mixes the feature statistics of two random instances in a single forward pass during training. The idea is grounded by the finding from recent style transfer research that feature statistics capture image style information, which essentially defines visual domains. Therefore, mixing feature statistics can be seen as an efficient way to synthesize new domains in the feature space, thus achieving data augmentation. MixStyle is easy to implement with a few lines of code, does not require modification to training objectives, and can fit a variety of learning paradigms including supervised domain generalization, semi-supervised domain generalization, and unsupervised domain adaptation. Our experiments show that MixStyle can significantly boost out-of-distribution generalization performance across a wide range of tasks including image recognition, instance retrieval and reinforcement learning. The source code is released at https://github.com/KaiyangZhou/mixstyle-release. [ABSTRACT FROM AUTHOR]

Deep neural networks (DNNs) are susceptible to adversarial examples generally generated by adding imperceptible perturbations to the clean images, resulting in the degraded performance of DNNs models. To generate adversarial examples, most methods utilize the Lp\documentclass[12pt]{minimal}\usepackage{amsmath}\usepackage{wasysym}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{amsbsy}\usepackage{mathrsfs}\usepackage{upgreek}\setlength{\oddsidemargin}{-69pt}\begin{document}$$L_p$$\end{document} norm to limit the perturbations and satisfy such imperceptibility. However, the Lp\documentclass[12pt]{minimal}\usepackage{amsmath}\usepackage{wasysym}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{amsbsy}\usepackage{mathrsfs}\usepackage{upgreek}\setlength{\oddsidemargin}{-69pt}\begin{document}$$L_p$$\end{document} norm cannot fully guarantee the semantic authenticity of adversarial examples. Defenses may take advantage of this defect to weaken the attack capability of adversarial examples. Moreover, existing methods with Lp\documentclass[12pt]{minimal}\usepackage{amsmath}\usepackage{wasysym}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{amsbsy}\usepackage{mathrsfs}\usepackage{upgreek}\setlength{\oddsidemargin}{-69pt}\begin{document}$$L_p$$\end{document} restriction have poor generalization ability in white-box attacks and have inferior aggressiveness in black-box attacks. To solve the problems mentioned above, we propose a multiple feature interpolation method to generate face adversarial examples. In the proposed method, we perform the multiple feature interpolation to generate face adversarial examples with new semantics in the process of original image reconstruction and conditional attribute-guided image generation based on StarGAN. Experimental results demonstrate that adversarial examples generated by our method possess new attribute-guided semantics and satisfactory attack success rates under both white-box and black-box settings. [ABSTRACT FROM AUTHOR]

From the n-BuOH extract of Broussonetia papyrifera seeds, a novel chushizilactam A ( 1) and adenosine ( 2) were isolated. Compound 1 was identified as 5ζ-hydroxy-5-(hydroxymethyl)-4-methyl-3-vinyl-1 H-pyrrol2(5 H)-one. Compounds 1 and 2 were evaluated for their cAMP-regulating activity by the AlphaScreen assay against N1E-115 cells. [ABSTRACT FROM AUTHOR]

Excessive proliferation and migration of vascular smooth muscle cells (VSMCs) contribute to the intimal hyperplasia in type 2 diabetes mellitus (T2DM) patients after percutaneous coronary intervention. We aimed to investigate the role of lncRNA cyclin-dependent kinase inhibitor 2B antisense RNA 1 (CDKN2B-AS1) in VSMC proliferation and migration, as well as the underlying mechanism. T2DM model mice with carotid balloon injury were used in vivo and mouse aortic vascular smooth muscle cells (MOVAS) stimulated by insulin were used in vitro to assess the role of CDKN2B-AS1 in VSMC proliferation and migration following vascular injury in T2DM state. To investigate cell viability and migration, MTT assay and Transwell assay were conducted. To elucidate the underlying molecular mechanisms, the methylation-specific polymerase chain reaction, RNA immunoprecipitation, RNA-pull down, co-immunoprecipitation, and chromatin immunoprecipitation were performed. In vivo, CDKN2B-AS1 was up-regulated in common carotid artery tissues. In vitro, insulin treatment increased CDKN2B-AS1 level, enhanced MOVAS cell proliferation and migration, while the promoting effect was reversed by CDKN2B-AS1 knockdown. CDKN2B-AS1 forms a complex with enhancer of zeste homolog 2 (EZH2) and DNA methyltransferase (cytosine-5) 1 (DNMT1) to regulate smooth muscle 22 alpha (SM22α) methylation levels. In insulin-stimulated cells, SM22α knockdown abrogated the inhibitory effect of CDKN2B-AS1 knockdown on cell viability and migration. Injection of lentivirus-sh-CDKN2B-AS1 relieved intimal hyperplasia in T2DM mice with carotid balloon injury. Up-regulation of CDKN2B-AS1 induced by insulin promotes cell proliferation and migration by targeting SM22α through forming a complex with EZH2 and DNMT1, thereby aggravating the intimal hyperplasia after vascular injury in T2DM. [ABSTRACT FROM AUTHOR]

The mechanism of unconventional superconductivity in correlated materials remains a great challenge in condensed matter physics. The recent discovery of superconductivity in infinite-layer nickelates, as an analog to high-Tc cuprates, has opened a new route to tackle this challenge. By growing 8 nm Pr0.8Sr0.2NiO2 films on the (LaAlO3)0.3(Sr2AlTaO6)0.7 substrate, we successfully raise the superconducting onset transition temperature Tc in the widely studied SrTiO3-substrated nickelates from 9 K into 15 K, which indicates compressive strain is an efficient protocol to further enhance superconductivity in infinite-layer nickelates. Additionally, the x-ray absorption spectroscopy, combined with the first-principles and many-body simulations, suggest a crucial role of the hybridization between Ni and O orbitals in the unconventional pairing. These results also suggest the increase of Tc be driven by the change of charge-transfer nature that would narrow the origin of general unconventional superconductivity in correlated materials to the covalence of transition metals and ligands. The recent discovery of superconductivity in the nickelates provides another angle to investigate this phenomenon in the high-Tc cuprates and hopefully help solve the mechanism of their unconventional superconductivity. Here, the authors report an increase in Tc for Pr0.8Sr0.2NiO2 where strain from the underlying LSAT substrate plays a possible role, supporting simulations also reveal the contributing role Ni and O orbitals hybridisation play in the unconventional pairing. [ABSTRACT FROM AUTHOR]

The coronavirus disease 2019 (COVID-19) brought a healthy crisis all around the world. It's not only affected healthcare utilization, but also some disease's incidence. We collected pre-hospital emergency data in Chengdu from January 2016 to December 2021, and studied the demand for emergency medical services (EMSs), emergency response times (ERTs), and the disease spectrum in the city proper of Chengdu as a whole. A total of 1,122,294 prehospital emergency medical service (EMS) instances met the inclusion criteria. Under the influence of COVID-19, notably in 2020, the epidemiological characteristics of prehospital emergency services in Chengdu were considerably altered. However, as the pandemic was brought under control, they returned to normal or even earlier in 2021.The COVID-19 pandemic had an effect on people's medical care-seeking behavior and the performance of prehospital emergency services. Although the indicators associated with prehospital emergency services eventually recovered as the epidemic was brought under control, they remained marginally different from before the outbreak. [ABSTRACT FROM AUTHOR]

Purpose: To investigate abnormalities in cortical and subcortical structures of the brain in preschool children with MRI-negative epilepsy. Methods: Cortical thickness, cortical mean curvature, cortical surface area, cortical volume, and volumes of subcortical structures were measured using Freesurfer software in preschool children with epilepsy and age-matched controls. Results: Findings showed cortical thickening in the left fusiform gyrus, left middle temporal gyrus, right suborbital sulcus, and right gyrus rectus, and cortical thinning mainly in the parietal lobe of preschool children with epilepsy compared to controls. The difference in cortical thickness in the left superior parietal lobule remained after correction for multiple comparisons and was negatively correlated with duration of epilepsy. Cortical mean curvature, surface area, and volume were mainly altered in the frontal and temporal lobes. Changes in mean curvature in the right pericallosal sulcus were positively correlated with age at seizure onset, and changes in mean curvature in the left intraparietal sulcus and transverse parietal sulcus were positively correlated with frequency of seizures. There were no significant differences in the volumes of the subcortical structures. Conclusion: Changes in preschool children with epilepsy occur in the cortical rather than subcortical structures of the brain. These findings further our understanding of the effects of epilepsy in preschool children and will inform management of epilepsy in this patient population. [ABSTRACT FROM AUTHOR]

A letter to the editor is presented in response to the article about lung cancer as the most common cancer-related death by Xiang-tao Ma, Li-wei Yu and Jing Fu.

Vision transformers (ViTs) have pushed the state-of-the-art for visual perception tasks. The self-attention mechanism underpinning the strength of ViTs has a quadratic complexity in both computation and memory usage. This motivates the development of approximating the self-attention at linear complexity. However, an in-depth analysis in this work reveals that existing methods are either theoretically flawed or empirically ineffective for visual recognition. We identify that their limitations are rooted in the inheritance of softmax based self-attention during approximations, that is, normalizing the scaled dot-product between token feature vectors using the softmax function. As preserving the softmax operation challenges any subsequent linearization efforts. By this insight, a family of SOftmax-Free Transformers (SOFT) are proposed. Specifically, a Gaussian kernel function is adopted to replace the dot-product similarity, enabling a full self-attention matrix to be approximated under low-rank matrix decomposition. For computational robustness, we estimate the Moore–Penrose inverse using an iterative Newton–Raphson method in the forward process only, while calculating its theoretical gradients only once in the backward process. To further expand applicability (e.g., dense prediction tasks), an efficient symmetric normalization technique is introduced. Extensive experiments on ImageNet, COCO and ADE20K show that our SOFT significantly improves the computational efficiency of existing ViT variants. With linear complexity, much longer token sequences are permitted by SOFT, resulting in superior trade-off between accuracy and complexity. Code and models are available at https://github.com/fudan-zvg/SOFT. [ABSTRACT FROM AUTHOR]

This paper addresses the problem of fully automated mining of public space video data, a highly desirable capability under contemporary commercial and security considerations. This task is especially challenging due to the complexity of the object behaviors to be profiled, the difficulty of analysis under the visual occlusions and ambiguities common in public space video, and the computational challenge of doing so in real-time. We address these issues by introducing a new dynamic topic model, termed a Markov Clustering Topic Model (MCTM). The MCTM builds on existing dynamic Bayesian network models and Bayesian topic models, and overcomes their drawbacks on sensitivity, robustness and efficiency. Specifically, our model profiles complex dynamic scenes by robustly clustering visual events into activities and these activities into global behaviours with temporal dynamics. A Gibbs sampler is derived for offline learning with unlabeled training data and a new approximation to online Bayesian inference is formulated to enable dynamic scene understanding and behaviour mining in new video data online in real-time. The strength of this model is demonstrated by unsupervised learning of dynamic scene models for four complex and crowded public scenes, and successful mining of behaviors and detection of salient events in each. [ABSTRACT FROM AUTHOR]

We propose a novel framework for automatic discovering and learning of behavioural context for video-based complex behaviour recognition and anomaly detection. Our work differs from most previous efforts on learning visual context in that our model learns multi-scale spatio-temporal rather than static context. Specifically three types of behavioural context are investigated: behaviour spatial context, behaviour correlation context, and behaviour temporal context. To that end, the proposed framework consists of an activity-based semantic scene segmentation model for learning behaviour spatial context, and a cascaded probabilistic topic model for learning both behaviour correlation context and behaviour temporal context at multiple scales. These behaviour context models are deployed for recognising non-exaggerated multi-object interactive and co-existence behaviours in public spaces. In particular, we develop a method for detecting subtle behavioural anomalies against the learned context. The effectiveness of the proposed approach is validated by extensive experiments carried out using data captured from complex and crowded outdoor scenes. [ABSTRACT FROM AUTHOR]

The purpose of this study was to explore the anatomical complexity and adjacent relationships of the sellar region in thin continuous sections so as to provide intimate morphological data for imaging diagnosis and surgical operations of the diseases in this region. After CT and MR examination verifying no brain lesions, one normal cadaver head was selected for this study from four Chinese adult male cadavers. After being embedded and frozen, the head was sliced into serial sections at 0.1 mm intervals in the transverse plane with SKC 500 computerized freezing milling machine. Then the serial transverse sections were photographed by a high-resolution digital camera and saved in the computer. Subsequently, the anatomic structures of the sellar region on the thin transverse sections were investigated and correlated with the MR images of the specimen as well as in vivo MR images, which were obtained from 20 normal Chinese male adult volunteers by a 3.0 T GE MR scanner. The base lines of the sectioning and the MR scan were all parallel to the AC-PC line. A total of 320 transverse sections and 10–12 transverse MR images related with the sellar region were obtained, respectively. We investigated the sectional anatomy of the sellar region and divided it into three parts: supra hypophysial area, hypophysial area and infra hypophysial area. The cavernous sinus was a venous passage full of blood and it could be divided into four interspaces according to its position relation with the internal carotid artery. The third, fourth, sixth cranial nerves and trigeminal branches ophthalmic nerve, maxillary nerve displayed from the anterior to the posterior in the lateral wall of cavernous sinus in transverse planes. Comparing continuous thin sections with MR images offers a better understanding of the complex anatomical structures and provides practical submillimeter anatomical data for imaging diagnosis and clinical treatment in this region. [ABSTRACT FROM AUTHOR]

The problem of fine-grained sketch-based image retrieval (FG-SBIR) is defined and investigated in this paper. In FG-SBIR, free-hand human sketch images are used as queries to retrieve photo images containing the same object instances. It is thus a cross-domain (sketch to photo) instance-level retrieval task. It is an extremely challenging problem because (i) visual comparisons and matching need to be executed under large domain gap, i.e., from black and white line drawing sketches to colour photos; (ii) it requires to capture the fine-grained (dis)similarities of sketches and photo images while free-hand sketches drawn by different people present different levels of deformation and expressive interpretation; and (iii) annotated cross-domain fine-grained SBIR datasets are scarce, challenging many state-of-the-art machine learning techniques, particularly those based on deep learning. In this paper, for the first time, we address all these challenges, providing a step towards the capabilities that would underpin a commercial sketch-based object instance retrieval application. Specifically, a new large-scale FG-SBIR database is introduced which is carefully designed to reflect the real-world application scenarios. A deep cross-domain matching model is then formulated to solve the intrinsic drawing style variability, large domain gap issues, and capture instance-level discriminative features. It distinguishes itself by a carefully designed attention module. Extensive experiments on the new dataset demonstrate the effectiveness of the proposed model and validate the need for a rigorous definition of the FG-SBIR problem and collecting suitable datasets. [ABSTRACT FROM AUTHOR]

Zero-shot learning (ZSL) aims to transfer knowledge from seen classes to unseen ones so that the latter can be recognised without any training samples. This is made possible by learning a projection function between a feature space and a semantic space (e.g. attribute space). Considering the seen and unseen classes as two domains, a big domain gap often exists which challenges ZSL. In this work, we propose a novel inductive ZSL model that leverages superclasses as the bridge between seen and unseen classes to narrow the domain gap. Specifically, we first build a class hierarchy of multiple superclass layers and a single class layer, where the superclasses are automatically generated by data-driven clustering over the semantic representations of all seen and unseen class names. We then exploit the superclasses from the class hierarchy to tackle the domain gap challenge in two aspects: deep feature learning and projection function learning. First, to narrow the domain gap in the feature space, we define a recurrent neural network over superclasses and then plug it into a convolutional neural network for enforcing the superclass hierarchy. Second, to further learn a transferrable projection function for ZSL, a novel projection function learning method is proposed by exploiting the superclasses to align the two domains. Importantly, our transferrable feature and projection learning methods can be easily extended to a closely related task—few-shot learning (FSL). Extensive experiments show that the proposed model outperforms the state-of-the-art alternatives in both ZSL and FSL tasks. [ABSTRACT FROM AUTHOR]

Orf virus (ORFV) infects sheep and goat tissues, resulting in severe proliferative lesions. To analyze cellular protein expression in ORFV-infected goat skin fibroblast (GSF) cells, we used two-dimensional liquid chromatography-tandem mass spectrometry coupled with isobaric tags for relative and absolute quantification (iTRAQ). The proteomics approach was used along with quantitative reverse transcription polymerase chain reaction (RT-qPCR) to detect differentially expressed proteins in ORFV-infected GSF cells and mock-infected GSF cells. A total of 282 differentially expressed proteins were identified. It was found that 222 host proteins were upregulated and 60 were downregulated following viral infection. We confirmed that these proteins were differentially expressed and found that heat shock 70-kDa protein 1B (HSPA1B) was differentially expressed and localized in the cytoplasm. It was also noted that HSPA1B caused inhibition of viral proliferation, in the middle and late stages of viral infection. The differentially expressed proteins were associated with the biological processes of viral binding, cell structure, signal transduction, cell adhesion, and cell proliferation. [ABSTRACT FROM AUTHOR]

In the era of big data remarked by high dimensionality and large sample size, the least absolute shrinkage and selection operator (LASSO) problems demand efficient algorithms. Both static and dynamic strategies based on screening test principle have been proposed recently, in order to safely filter out irrelevant atoms from the dictionary. However, such strategies only work well for LASSO problems with large regularization parameters, and lose their efficiency for those with small regularization parameters. This paper presents a novel greedy screening test strategy to accelerate solving LASSO problems with small regularization parameters, as well as its effectiveness through adoption of a relatively larger regularization parameter which filters out irrelevant atoms in every iteration. Further more, the convergence proof of the greedy strategy is given, and the computational complexity of LASSO solvers integrated with this strategy is investigated. Numerical experiments on both synthetic and real data sets support the effectiveness of this greedy strategy, and the results show it outperforms both the static and dynamic strategies for LASSO problems with small regularization parameters. [ABSTRACT FROM AUTHOR]

Purpose: This postmortem magnetic resonance imaging (MRI) study of the fetal spine aimed to describe the timing of appearance, shape, volume, and relative positions of the S1-S3 costal element ossification centers (CEOCs).Methods: We obtained sagittal 3D dual-echo steady-state with water excitation T2 images of the entire spine in 71 fetuses (gestational ages (GAs), 17-42 weeks). Computed tomography and histological examinations were performed on two fetal specimens (GAs, 21 and 30 weeks) to validate the MR images. The presence/absence of each sacral CEOC was recorded according to the GA. CEOC volume was measured. We analyzed the CEOC position relative to the vertebral column and ilium.Results: The S1, S2, and S3 CEOCs first appeared at 23, 22, and 29 weeks, respectively. The S1 and S2 CEOCs could be detected in all fetuses with GAs of ≥ 30 weeks and ≥ 35 weeks, respectively, while the S3 CEOCs were variably present until term. The percentages of detection of the S1 and S2 CEOCs were significantly greater than that of the S3 CEOCs at each GA. At S1 and S2, the CEOC volume increased exponentially with GA. The relative positions of the S1 and S2 CEOCs, but not the S3 CEOCs, significantly correlated with GA (P < 0.001).Conclusion: We have described the timeline of appearance as well as the volume and position of the S1-S3 CEOCs in the fetal spine on postmortem MRI according to GA. [ABSTRACT FROM AUTHOR]

Most existing person re-identification (re-id) methods are unsuitable for real-world deployment due to two reasons: Unscalability to large population size, and Inadaptability over time. In this work, we present a unified solution to address both problems. Specifically, we propose to construct an identity regression space (IRS) based on embedding different training person identities (classes) and formulate re-id as a regression problem solved by identity regression in the IRS. The IRS approach is characterised by a closed-form solution with high learning efficiency and an inherent incremental learning capability with human-in-the-loop. Extensive experiments on four benchmarking datasets (VIPeR, CUHK01, CUHK03 and Market-1501) show that the IRS model not only outperforms state-of-the-art re-id methods, but also is more scalable to large re-id population size by rapidly updating model and actively selecting informative samples with reduced human labelling effort. [ABSTRACT FROM AUTHOR]

Purpose: To describe the temporal pattern of the appearance of the S1-Co1 centrum ossification centers (COCs) and provide reference data for the S1-S5 COCs and sacral length at various gestational ages (GAs).Methods: Postmortem magnetic resonance imaging (MRI) was performed on 71 fetuses (GA, 17-42 weeks) using the 3D dual-echo steady-state with water excitation T2 sequence in the sagittal plane. To confirm the reliability of this sequence, the MRI data were compared with the CT and histologic data obtained from two fetuses (GAs, 21 and 30 weeks). The presence or absence of each sacrococcygeal COC was recorded. Sacral length and S1-S5 COC height, sagittal diameter, transverse diameter, cross-sectional area, and volume were measured.Results: All fetuses showed S1-S3 COCs by 17 weeks, S4 COCs by 19 weeks, and S5 COCs by 28 weeks. The S4, S5, and Co-1 COCs were visualized in 70 (98.59%), 51 (71.83%), and 21 (29.58%) fetuses, respectively. Sacral length, height, sagittal, and transverse diameters increased linearly, while cross-sectional area and volume increased exponentially with advancing GA. Mean growth rates of the sagittal and transverse diameters, cross-sectional area, and volume, but not of height, significantly differed among the S1-S5 vertebrae.Conclusion: We have presented the timing of appearance of individual sacrococcygeal COCs and the age-specific, normative MRI reference values for sacral length and the morphometric parameters of the sacral COCs, which are of clinical importance in the diagnosis of congenital sacral abnormalities and skeletal dysplasia. [ABSTRACT FROM AUTHOR]

Mechanical bonded SUS304/Q235B bimetallic pipes have been widely used as subsea oil and gas transmission pipeline. Due to the complexity of the subsea environment, the thin inner layer of SUS304 might crinkle and swell locally and then separated from the base layer of Q235B. In this work, lap welding inside the SUS304/Q235B bimetallic pipes was investigated with special emphasis on the corrosion resistance of the near surface region of the lap joint. Nothing the significant difference of weld width between its upper part and lower part of typical laser welded “wine-glass” type bead, laser welding was employed here to reduce the dilution of alloy elements in the fusion zone of inner stainless steel layer. The results showed that content of major alloy elements such as chromium (Cr) and nickel (Ni) in the near surface region of fusion zone could be kept as about 14 and 6 wt.%, respectively, by optimizing the laser lap welding process parameters. The laser welded SUS304/Q235B overlap joints had excellent shear strength, and the shear fracture was consisted of numerous dimples. Finally, laser filler wire cladding of SUS316 stainless steel on the top surface of previously achieved overlap joint was carried out to ensure that upper part of the weld zone had enough corrosion resistance. [ABSTRACT FROM AUTHOR]

We propose a deep learning approach to free-hand sketch recognition that achieves state-of-the-art performance, significantly surpassing that of humans. Our superior performance is a result of modelling and exploiting the unique characteristics of free-hand sketches, i.e., consisting of an ordered set of strokes but lacking visual cues such as colour and texture, being highly iconic and abstract, and exhibiting extremely large appearance variations due to different levels of abstraction and deformation. Specifically, our deep neural network, termed Sketch-a-Net has the following novel components: (i) we propose a network architecture designed for sketch rather than natural photo statistics. (ii) Two novel data augmentation strategies are developed which exploit the unique sketch-domain properties to modify and synthesise sketch training data at multiple abstraction levels. Based on this idea we are able to both significantly increase the volume and diversity of sketches for training, and address the challenge of varying levels of sketching detail commonplace in free-hand sketches. (iii) We explore different network ensemble fusion strategies, including a re-purposed joint Bayesian scheme, to further improve recognition performance. We show that state-of-the-art deep networks specifically engineered for photos of natural objects fail to perform well on sketch recognition, regardless whether they are trained using photos or sketches. Furthermore, through visualising the learned filters, we offer useful insights in to where the superior performance of our network comes from. [ABSTRACT FROM AUTHOR]

Analyzing unstructured group activities and events in uncontrolled web videos is a challenging task due to (1) the semantic gap between class labels and low-level visual features, (2) the demanding computational cost given high-dimensional low-level feature vectors and (3) the lack of labeled training data. These difficulties can be overcome by learning a meaningful and compact mid-level video representation. To this end, in this paper a novel supervised probabilistic graphical model termed Relevance Restricted Boltzmann Machine (ReRBM) is developed to learn a low-dimensional latent semantic representation for complex activities and events. Our model is a variant of the Restricted Boltzmann Machine (RBM) with a number of critical extensions: (1) sparse Bayesian learning is incorporated into the RBM to learn features which are relevant to video classes, i.e., discriminative; (2) binary stochastic hidden units in the RBM are replaced by rectified linear units in order to better explain complex video contents and make variational inference tractable for the proposed model; and (3) an efficient variational EM algorithm is formulated for model parameter estimation and inference. We conduct extensive experiments on two recent challenging benchmarks: the Unstructured Social Activity Attribute dataset and the Event Video dataset. The experimental results demonstrate that the relevant features learned by our model provide better semantic and discriminative description for videos than a number of alternative supervised latent variable models, and achieves state of the art performance in terms of classification accuracy and retrieval precision, particularly when only a few labeled training samples are available. [ABSTRACT FROM AUTHOR]

Canine parvovirus (CPV) can cause acute hemorrhagic diarrhea and fatal myocarditis in young dogs. Currently, most studies have focused on the evolution of the VP2 gene, whereas the full-length genome of CPV has been rarely reported. In this study, the whole genomes of CPV-LZ1 and CPV-LZ2 strains prevalent in Northwest China were determined and analyzed in comparison with those of the reference CPVs. The genome sequences of both LZ strains consisted of 5053 nucleotides. CPV-LZ1 and CPV-LZ2 strains were designated as new CPV-2a and CPV-2b, respectively. Sequence alignment analysis results revealed that these two new strains underwent specific unique variations during the process of local adaption. The left non-translated regions of these strains formed a Y-shaped hairpin structure, whereas the right non-translated regions lacked the reiteration of DNA sequence. A phylogenetic tree constructed from 33 whole coding regions of CPVs showed a strong spatial clustering, and these two strains belonged to the Chinese strain cluster lineage. This study provides a method to obtain the full-length genome of CPV. The isolation and characterization of these viruses adds incrementally to the knowledge of the full-length genome of CPV. The results from this study also provide insight into the molecular epidemiology and genetic diversity of the CPV field isolates from Northwest China and can be useful in preventing and controlling CPV infection in this region. [ABSTRACT FROM AUTHOR]

A letter to the editor is presented in response to the article "Patellar resurfacing versus nonresurfacing in total knee arthroplasty: A meta-analysis of randomised controlled trials" by K. Chen and colleagues in a previous issue of the journal.

The record superconducting transition temperature (Tc) for the iron-based high-temperature superconductors (Fe-HTS) has long been 56 K. Recently, in single-layer FeSe films grown on SrTiO3 substrates, indications of a new record of 65 K have been reported. Using in situ photoemission measurements, we substantiate the presence of spin density waves (SDWs) in FeSe films-a key ingredient of Fe-HTS that was missed in FeSe before-and we find that this weakens with increased thickness or reduced strain. We demonstrate that the superconductivity occurs when the electrons transferred from the oxygen-vacant substrate suppress the otherwise pronounced SDWs in single-layer FeSe. Beyond providing a comprehensive understanding of FeSe films and directions to further enhance its Tc, we map out the phase diagram of FeSe as a function of lattice constant, which contains all the essential physics of Fe-HTS. With the simplest structure, cleanest composition and single tuning parameter, monolayer FeSe is an ideal system for testing theories of Fe-HTS. [ABSTRACT FROM AUTHOR]

Elemental sulfur bio-oxidation by the typical acidophilic sulfur-oxidizing microbe Acidithiobacillus ferrooxidans was investigated by using the technique of sulfur K-edge XANES spectroscopy. Our results showed that the majority of elemental sulfur altered by A. ferrooxidans was dissolved into the organic phase containing carbon disulfide, while a part of it floated. The fitted results of sulfur K-edge XANES spectrum of the floated sulfur showed that the floating part of the elemental sulfur powder was converted to polymeric sulfur and the relative concentration of sulfur in cyclo-octasulfur S and polymeric sulfur was 37.2 and 62.8%, respectively. It seems that the cyclo-octasulfur is converted to the polymeric sulfur and this appears to be necessary for oxidation of elemental sulfur by A. ferrooxidans. The results have important implications for our understanding of the mechanisms for bio-oxidation of elemental sulfur. [ABSTRACT FROM AUTHOR]

Poultry meat quality is associated with breed, age, tissue and other factors. Many previous studies have focused on distinct breeds; however, little is known regarding the epigenetic regulatory mechanisms in different age stages, such as DNA methylation. Here, we compared the global DNA methylation profiles between juvenile (20 weeks old) and later laying-period (55 weeks old) hens and identified candidate genes related to the development and meat quality of breast muscle using whole-genome bisulfite sequencing. The results showed that the later laying-period hens, which had a higher intramuscular fat (IMF) deposition capacity and water holding capacity (WHC) and less tenderness, exhibited higher global DNA methylation levels than the juvenile hens. A total of 2,714 differentially methylated regions were identified in the present study, which corresponded to 378 differentially methylated genes, mainly affecting muscle development, lipid metabolism, and the ageing process. Hypermethylation of the promoters of the genes ABCA1, COL6A1 and GSTT1L and the resulting transcriptional down-regulation in the later laying-period hens may be the reason for the significant difference in the meat quality between the juvenile and later laying-period hens. These findings contribute to a better understanding of epigenetic regulation in the skeletal muscle development and meat quality of chicken. [ABSTRACT FROM AUTHOR]