Abstract: BackgroundPatients with ischemic stroke or transient ischemic attack (TIA) are at increased risk for future cardiovascular events despite current preventive therapies. The identification of insulin resistance as a risk factor for stroke and myocardial infarction raised the possibility that pioglitazone, which improves insulin sensitivity, might benefit patients with cerebrovascular disease. MethodsIn this multicenter, double-blind trial, we randomly assigned 3876 patients who had had a recent ischemic stroke or TIA to receive either pioglitazone (target dose, 45 mg daily) or placebo. Eligible patients did not have diabetes but were found to have insulin resistance on the basis of a score of more than 3.0 on the homeostasis model assessment of insulin resistance (HOMA-IR) index. The primary outcome was fatal or nonfatal stroke or myocardial infarction. ResultsBy 4.8 years, a primary outcome had occurred in 175 of 1939 patients (9.0%) in the pioglitazone group and in 228 of 1937 (11.8%) in the placebo group...

Abstract: In recent years, indoor positioning has emerged as a critical function in many end-user applications; including military, civilian, disaster relief and peacekeeping missions. In comparison with outdoor environments, sensing location information in indoor environments requires a higher precision and is a more challenging task in part because various objects reflect and disperse signals. Ultra WideBand (UWB) is an emerging technology in the field of indoor positioning that has shown better performance compared to others. In order to set the stage for this work, we provide a survey of the state-of-the-art technologies in indoor positioning, followed by a detailed comparative analysis of UWB positioning technologies. We also provide an analysis of strengths, weaknesses, opportunities, and threats (SWOT) to analyze the present state of UWB positioning technologies. While SWOT is not a quantitative approach, it helps in assessing the real status and in revealing the potential of UWB positioning to effectively address the indoor positioning problem. Unlike previous studies, this paper presents new taxonomies, reviews some major recent advances, and argues for further exploration by the research community of this challenging problem space.

Abstract: The chemogenetic technology DREADD (designer receptors exclusively activated by designer drugs) is widely used for remote manipulation of neuronal activity in freely moving animals. DREADD technology posits the use of “designer receptors,” which are exclusively activated by the “designer drug” clozapine N-oxide (CNO). Nevertheless, the in vivo mechanism of action of CNO at DREADDs has never been confirmed. CNO does not enter the brain after systemic drug injections and shows low affinity for DREADDs. Clozapine, to which CNO rapidly converts in vivo, shows high DREADD affinity and potency. Upon systemic CNO injections, converted clozapine readily enters the brain and occupies central nervous system–expressed DREADDs, whereas systemic subthreshold clozapine injections induce preferential DREADD-mediated behaviors.

Abstract: Previous studies have shown that the method of hydrogen mass repartitioning (HMR) is a potentially useful tool for accelerating molecular dynamics (MD) simulations. By repartitioning the mass of heavy atoms into the bonded hydrogen atoms, it is possible to slow the highest-frequency motions of the macromolecule under study, thus allowing the time step of the simulation to be increased by up to a factor of 2. In this communication, we investigate further how this mass repartitioning allows the simulation time step to be increased in a stable fashion without significantly increasing discretization error. To this end, we ran a set of simulations with different time steps and mass distributions on a three-residue peptide to get a comprehensive view of the effect of mass repartitioning and time step increase on a system whose accessible phase space is fully explored in a relatively short amount of time. We next studied a 129-residue protein, hen egg white lysozyme (HEWL), to verify that the observed behavior extends to a larger, more-realistic, system. Results for the protein include structural comparisons from MD trajectories, as well as comparisons of pKa calculations via constant-pH MD. We also calculated a potential of mean force (PMF) of a dihedral rotation for the MTS [(1-oxyl-2,2,5,5-tetramethyl-pyrroline-3-methyl)methanethiosulfonate] spin label via umbrella sampling with a set of regular MD trajectories, as well as a set of mass-repartitioned trajectories with a time step of 4 fs. Since no significant difference in kinetics or thermodynamics is observed by the use of fast HMR trajectories, further evidence is provided that long-time-step HMR MD simulations are a viable tool for accelerating MD simulations for molecules of biochemical interest.

Abstract: The merger of two neutron stars is predicted to give rise to three major detectable phenomena: a short burst of gamma-rays, a gravitational wave signal, and a transient optical/near-infrared source powered by the synthesis of large amounts of very heavy elements via rapid neutron capture (the r-process). Such transients, named "macronovae" or "kilonovae", are believed to be centres of production of rare elements such as gold and platinum. The most compelling evidence so far for a kilonova was a very faint near-infrared rebrightening in the afterglow of a short gamma-ray burst at z = 0.356, although findings indicating bluer events have been reported. Here we report the spectral identification and describe the physical properties of a bright kilonova associated with the gravitational wave source GW 170817 and gamma-ray burst GRB 170817A associated with a galaxy at a distance of 40 Mpc from Earth. Using a series of spectra from ground-based observatories covering the wavelength range from the ultraviolet to the near-infrared, we find that the kilonova is characterized by rapidly expanding ejecta with spectral features similar to those predicted by current models. The ejecta is optically thick early on, with a velocity of about 0.2 times light speed, and reaches a radius of about 50 astronomical units in only 1.5 days. As the ejecta expands, broad absorption-like lines appear on the spectral continuum indicating atomic species produced by nucleosynthesis that occurs in the post-merger fast-moving dynamical ejecta and in two slower (0.05 times light speed) wind regions. Comparison with spectral models suggests that the merger ejected 0.03-0.05 solar masses of material, including high-opacity lanthanides.

Abstract: In this paper we discuss several forms of spatiotemporal convolutions for video analysis and study their effects on action recognition. Our motivation stems from the observation that 2D CNNs applied to individual frames of the video have remained solid performers in action recognition. In this work we empirically demonstrate the accuracy advantages of 3D CNNs over 2D CNNs within the framework of residual learning. Furthermore, we show that factorizing the 3D convolutional filters into separate spatial and temporal components yields significantly advantages in accuracy. Our empirical study leads to the design of a new spatiotemporal convolutional block "R(2+1)D" which gives rise to CNNs that achieve results comparable or superior to the state-of-the-art on Sports-1M, Kinetics, UCF101 and HMDB51.

Abstract: SUMMARYEpigenetic changes are present in all human cancers and are now known to cooperate with genetic alterations to drive the cancer phenotype. These changes involve DNA methylation, histone modifiers and readers, chromatin remodelers, microRNAs, and other components of chromatin. Cancer genetics and epigenetics are inextricably linked in generating the malignant phenotype; epigenetic changes can cause mutations in genes, and, conversely, mutations are frequently observed in genes that modify the epigenome. Epigenetic therapies, in which the goal is to reverse these changes, are now one standard of care for a preleukemic disorder and form of lymphoma. The application of epigenetic therapies in the treatment of solid tumors is also emerging as a viable therapeutic route.

Abstract: With the aim of developing a fast yet accurate algorithm for compressive sensing (CS) reconstruction of natural images, we combine in this paper the merits of two existing categories of CS methods: the structure insights of traditional optimization-based methods and the speed of recent network-based ones. Specifically, we propose a novel structured deep network, dubbed ISTA-Net, which is inspired by the Iterative Shrinkage-Thresholding Algorithm (ISTA) for optimizing a general $$ norm CS reconstruction model. To cast ISTA into deep network form, we develop an effective strategy to solve the proximal mapping associated with the sparsity-inducing regularizer using nonlinear transforms. All the parameters in ISTA-Net (e.g. nonlinear transforms, shrinkage thresholds, step sizes, etc.) are learned end-to-end, rather than being hand-crafted. Moreover, considering that the residuals of natural images are more compressible, an enhanced version of ISTA-Net in the residual domain, dubbed ISTA-Net+, is derived to further improve CS reconstruction. Extensive CS experiments demonstrate that the proposed ISTA-Nets outperform existing state-of-the-art optimization-based and network-based CS methods by large margins, while maintaining fast computational speed. Our source codes are available: http://jianzhang.tech/projects/ISTA-Net.

Abstract: SARS-CoV-2 outbreak is the first pandemic of the century. SARS-CoV-2 infection is transmitted through droplets; other transmission routes are hypothesized but not confirmed. So far, it is unclear whether and how SARS-CoV-2 can be transmitted from the mother to the fetus. We demonstrate the transplacental transmission of SARS-CoV-2 in a neonate born to a mother infected in the last trimester and presenting with neurological compromise. The transmission is confirmed by comprehensive virological and pathological investigations. In detail, SARS-CoV-2 causes: (1) maternal viremia, (2) placental infection demonstrated by immunohistochemistry and very high viral load; placental inflammation, as shown by histological examination and immunohistochemistry, and (3) neonatal viremia following placental infection. The neonate is studied clinically, through imaging, and followed up. The neonate presented with neurological manifestations, similar to those described in adult patients.

Abstract: CRISPR/Cas systems confer immunity against invading nucleic acids and phages in bacteria and archaea. CRISPR/Cas13a (known previously as C2c2) is a class 2 type VI-A ribonuclease capable of targeting and cleaving single-stranded RNA (ssRNA) molecules of the phage genome. Here, we employ CRISPR/Cas13a to engineer interference with an RNA virus, Turnip Mosaic Virus (TuMV), in plants. CRISPR/Cas13a produces interference against green fluorescent protein (GFP)-expressing TuMV in transient assays and stable overexpression lines of Nicotiana benthamiana. CRISPR RNA (crRNAs) targeting the HC-Pro and GFP sequences exhibit better interference than those targeting other regions such as coat protein (CP) sequence. Cas13a can also process pre-crRNAs into functional crRNAs. Our data indicate that CRISPR/Cas13a can be used for engineering interference against RNA viruses, providing a potential novel mechanism for RNA-guided immunity against RNA viruses and for other RNA manipulations in plants.

Abstract: This study investigated user acceptance, concerns, and willingness to buy partially, highly, and fully automated vehicles. By means of a 63-question Internet-based survey, we collected 5000 responses from 109 countries (40 countries with at least 25 respondents). We determined cross-national differences, and assessed correlations with personal variables, such as age, gender, and personality traits as measured with a short version of the Big Five Inventory. Results showed that respondents, on average, found manual driving the most enjoyable mode of driving. Responses were diverse: 22% of the respondents did not want to pay more than $0 for a fully automated driving system, whereas 5% indicated they would be willing to pay more than $30,000, and 33% indicated that fully automated driving would be highly enjoyable. 69% of respondents estimated that fully automated driving will reach a 50% market share between now and 2050. Respondents were found to be most concerned about software hacking/misuse, and were also concerned about legal issues and safety. Respondents scoring higher on neuroticism were slightly less comfortable about data transmitting, whereas respondents scoring higher on agreeableness were slightly more comfortable with this. Respondents from more developed countries (in terms of lower accident statistics, higher education, and higher income) were less comfortable with their vehicle transmitting data, with cross-national correlations between ρ = −0.80 and ρ = −0.90. The present results indicate the major areas of promise and concern among the international public, and could be useful for vehicle developers and other stakeholders.

Abstract: Strongly positive temperature anomalies developed in the NE Pacific Ocean during the boreal winter of 2013–2014. Based on a mixed layer temperature budget, these anomalies were caused by lower than normal rates of the loss of heat from the ocean to the atmosphere and of relatively weak cold advection in the upper ocean. Both of these mechanisms can be attributed to an unusually strong and persistent weather pattern featuring much higher than normal sea level pressure over the waters of interest. This anomaly was the greatest observed in this region since at least the 1980s. The region of warm sea surface temperature anomalies subsequently expanded and reached coastal waters in spring and summer 2014. Impacts on fisheries and regional weather are discussed. It is found that sea surface temperature anomalies in this region affect air temperatures downwind in Washington state.

Abstract: . The global methane (CH4) budget is becoming an increasingly important component for managing realistic pathways to mitigate climate change. This relevance, due to a shorter atmospheric lifetime and a stronger warming potential than carbon dioxide, is challenged by the still unexplained changes of atmospheric CH4 over the past decade. Emissions and concentrations of CH4 are continuing to increase, making CH4 the second most important human-induced greenhouse gas after carbon dioxide. Two major difficulties in reducing uncertainties come from the large variety of diffusive CH4 sources that overlap geographically, and from the destruction of CH4 by the very short-lived hydroxyl radical (OH). To address these difficulties, we have established a consortium of multi-disciplinary scientists under the umbrella of the Global Carbon Project to synthesize and stimulate research on the methane cycle, and producing regular (∼ biennial) updates of the global methane budget. This consortium includes atmospheric physicists and chemists, biogeochemists of surface and marine emissions, and socio-economists who study anthropogenic emissions. Following Kirschke et al. (2013), we propose here the first version of a living review paper that integrates results of top-down studies (exploiting atmospheric observations within an atmospheric inverse-modelling framework) and bottom-up models, inventories and data-driven approaches (including process-based models for estimating land surface emissions and atmospheric chemistry, and inventories for anthropogenic emissions, data-driven extrapolations). For the 2003–2012 decade, global methane emissions are estimated by top-down inversions at 558 Tg CH4 yr−1, range 540–568. About 60 % of global emissions are anthropogenic (range 50–65 %). Since 2010, the bottom-up global emission inventories have been closer to methane emissions in the most carbon-intensive Representative Concentrations Pathway (RCP8.5) and higher than all other RCP scenarios. Bottom-up approaches suggest larger global emissions (736 Tg CH4 yr−1, range 596–884) mostly because of larger natural emissions from individual sources such as inland waters, natural wetlands and geological sources. Considering the atmospheric constraints on the top-down budget, it is likely that some of the individual emissions reported by the bottom-up approaches are overestimated, leading to too large global emissions. Latitudinal data from top-down emissions indicate a predominance of tropical emissions (∼ 64 % of the global budget, The most important source of uncertainty on the methane budget is attributable to emissions from wetland and other inland waters. We show that the wetland extent could contribute 30–40 % on the estimated range for wetland emissions. Other priorities for improving the methane budget include the following: (i) the development of process-based models for inland-water emissions, (ii) the intensification of methane observations at local scale (flux measurements) to constrain bottom-up land surface models, and at regional scale (surface networks and satellites) to constrain top-down inversions, (iii) improvements in the estimation of atmospheric loss by OH, and (iv) improvements of the transport models integrated in top-down inversions. The data presented here can be downloaded from the Carbon Dioxide Information Analysis Center ( http://doi.org/10.3334/CDIAC/GLOBAL_METHANE_BUDGET_2016_V1.1 ) and the Global Carbon Project.

Abstract: Earlier research primarily attributed the effects of mesenchymal stem cell (MSC) therapies to their capacity for local engrafting and differentiating into multiple tissue types. However, recent studies have revealed that implanted cells do not survive for long, and that the benefits of MSC therapy could be due to the vast array of bioactive factors they produce, which play an important role in the regulation of key biologic processes. Secretome derivatives, such as conditioned media or exosomes, may present considerable advantages over cells for manufacturing, storage, handling, product shelf life and their potential as a ready-to-go biologic product. Nevertheless, regulatory requirements for manufacturing and quality control will be necessary to establish the safety and efficacy profile of these products. Among MSCs, human uterine cervical stem cells (hUCESCs) may be a good candidate for obtaining secretome-derived products. hUCESCs are obtained by Pap cervical smear, which is a less invasive and painful method than those used for obtaining other MSCs (for example, from bone marrow or adipose tissue). Moreover, due to easy isolation and a high proliferative rate, it is possible to obtain large amounts of hUCESCs or secretome-derived products for research and clinical use.

Abstract: The novel coronavirus disease (COVID-19) has been rapidly transmitted in China, Macau, Hong Kong, and other Asian and European counterparts. This COVID-19 epidemic has aroused increasing attention nationwide. Patients, health professionals, and the general public are under insurmountable psychological pressure which may lead to various psychological problems, such as anxiety, fear, depression, and insomnia. Psychological crisis intervention plays a pivotal role in the overall deployment of the disease control. The National Health Commission of China has summoned a call for emergency psychological crisis intervention and thus, various mental health associations and organizations have established expert teams to compile guidelines and public health educational articles/videos for mental health professionals and the general public alongside with online mental health services. In addition, mental health professionals and expert groups are stationed in designated isolation hospitals to provide on-site services. Experts have reached a consensus on the admission of patients with severe mental illness during the COVID-19 outbreak in mental health institutions. Nevertheless, the rapid transmission of the COVID-19 has emerged to mount a serious challenge to the mental health service in China.

Abstract: This paper demonstrates how to use the R package stm for structural topic modeling. The structural topic model allows researchers to flexibly estimate a topic model that includes document-level metadata. Estimation is accomplished through a fast variational approximation. The stm package provides many useful features, including rich ways to explore topics, estimate uncertainty, and visualize quantities of interest.

Abstract: Recent work on automated augmentation strategies has led to state-of-the-art results in image classification and object detection. An obstacle to a large-scale adoption of these methods is that they require a separate and expensive search phase. A common way to overcome the expense of the search phase was to use a smaller proxy task. However, it was not clear if the optimized hyperparameters found on the proxy task are also optimal for the actual task. In this work, we rethink the process of designing automated augmentation strategies. We find that while previous work required a search for both magnitude and probability of each operation independently, it is sufficient to only search for a single distortion magnitude that jointly controls all operations. We hence propose a simplified search space that vastly reduces the computational expense of automated augmentation, and permits the removal of a separate proxy task. Despite the simplifications, our method achieves equal or better performance over previous automated augmentation strategies on on CIFAR-10/100, SVHN, ImageNet and COCO datasets. EfficientNet-B7, we achieve 85.0% accuracy, a 1.0% increase over baseline augmentation, a 0.6% improvement over AutoAugment on the ImageNet dataset. With EfficientNet-B8, we achieve 85.4% accuracy on ImageNet, which matches a previous result that used 3.5B extra images. On object detection, the same method as classification leads to 1.0-1.3% improvement over baseline augmentation. Code will be made available online.

Abstract: The focus of this review is to provide an overview of the chitosan based nanoparticles for various non-parenteral applications and also to put a spotlight on current research including sustained release and mucoadhesive chitosan dosage forms. Chitosan is a biodegradable, biocompatible polymer regarded as safe for human dietary use and approved for wound dressing applications. Chitosan has been used as a carrier in polymeric nanoparticles for drug delivery through various routes of administration. Chitosan has chemical functional groups that can be modified to achieve specific goals, making it a polymer with a tremendous range of potential applications. Nanoparticles (NP) prepared with chitosan and chitosan derivatives typically possess a positive surface charge and mucoadhesive properties such that can adhere to mucus membranes and release the drug payload in a sustained release manner. Chitosan-based NP have various applications in non-parenteral drug delivery for the treatment of cancer, gastrointestinal diseases, pulmonary diseases, drug delivery to the brain and ocular infections which will be exemplified in this review. Chitosan shows low toxicity both in vitro and some in vivo models. This review explores recent research on chitosan based NP for non-parenteral drug delivery, chitosan properties, modification, toxicity, pharmacokinetics and preclinical studies.

Abstract: The global urgency to improve STEM education may be driven by environmental and social impacts of the twenty-first century which in turn jeopardizes global security and economic stability. The complexity of these global factors reach beyond just helping students achieve high scores in math and science assessments. Friedman (The world is flat: A brief history of the twenty-first century, 2005) helped illustrate the complexity of a global society, and educators must help students prepare for this global shift. In response to these challenges, the USA experienced massive STEM educational reforms in the last two decades. In practice, STEM educators lack cohesive understanding of STEM education. Therefore, they could benefit from a STEM education conceptual framework. The process of integrating science, technology, engineering, and mathematics in authentic contexts can be as complex as the global challenges that demand a new generation of STEM experts. Educational researchers indicate that teachers struggle to make connections across the STEM disciplines. Consequently, students are often disinterested in science and math when they learn in an isolated and disjoined manner missing connections to crosscutting concepts and real-world applications. The following paper will operationalize STEM education key concepts and blend learning theories to build an integrated STEM education framework to assist in further researching integrated STEM education.

Abstract: Cheminformaticians are equipped with a very rich toolbox when carrying out molecular similarity calculations. A large number of molecular representations exist, and there are several methods (similarity and distance metrics) to quantify the similarity of molecular representations. In this work, eight well-known similarity/distance metrics are compared on a large dataset of molecular fingerprints with sum of ranking differences (SRD) and ANOVA analysis. The effects of molecular size, selection methods and data pretreatment methods on the outcome of the comparison are also assessed. A supplier database ( https://mcule.com/ ) was used as the source of compounds for the similarity calculations in this study. A large number of datasets, each consisting of one hundred compounds, were compiled, molecular fingerprints were generated and similarity values between a randomly chosen reference compound and the rest were calculated for each dataset. Similarity metrics were compared based on their ranking of the compounds within one experiment (one dataset) using sum of ranking differences (SRD), while the results of the entire set of experiments were summarized on box and whisker plots. Finally, the effects of various factors (data pretreatment, molecule size, selection method) were evaluated with analysis of variance (ANOVA). This study complements previous efforts to examine and rank various metrics for molecular similarity calculations. Here, however, an entirely general approach was taken to neglect any a priori knowledge on the compounds involved, as well as any bias introduced by examining only one or a few specific scenarios. The Tanimoto index, Dice index, Cosine coefficient and Soergel distance were identified to be the best (and in some sense equivalent) metrics for similarity calculations, i.e. these metrics could produce the rankings closest to the composite (average) ranking of the eight metrics. The similarity metrics derived from Euclidean and Manhattan distances are not recommended on their own, although their variability and diversity from other similarity metrics might be advantageous in certain cases (e.g. for data fusion). Conclusions are also drawn regarding the effects of molecule size, selection method and data pretreatment on the ranking behavior of the studied metrics.

Abstract: Traditional1 salient object detection models often use hand-crafted features to formulate contrast and various prior knowledge, and then combine them artificially. In this work, we propose a novel end-to-end deep hierarchical saliency network (DHSNet) based on convolutional neural networks for detecting salient objects. DHSNet first makes a coarse global prediction by automatically learning various global structured saliency cues, including global contrast, objectness, compactness, and their optimal combination. Then a novel hierarchical recurrent convolutional neural network (HRCNN) is adopted to further hierarchically and progressively refine the details of saliency maps step by step via integrating local context information. The whole architecture works in a global to local and coarse to fine manner. DHSNet is directly trained using whole images and corresponding ground truth saliency masks. When testing, saliency maps can be generated by directly and efficiently feedforwarding testing images through the network, without relying on any other techniques. Evaluations on four benchmark datasets and comparisons with other 11 state-of-the-art algorithms demonstrate that DHSNet not only shows its significant superiority in terms of performance, but also achieves a real-time speed of 23 FPS on modern GPUs.

Abstract: Autophagy is a conserved catabolic process that degrades cytoplasmic constituents and organelles in the lysosome. Starvation-induced protein degradation is a salient feature of autophagy but recent progress has illuminated how autophagy, during both starvation and nutrient-replete conditions, can mobilize diverse cellular energy and nutrient stores such as lipids, carbohydrates and iron. Processes such as lipophagy, glycophagy and ferritinophagy enable cells to salvage key metabolites to sustain and facilitate core anabolic functions. Here, we discuss the established and emerging roles of autophagy in fuelling biosynthetic capacity and in promoting metabolic and nutrient homeostasis.

Abstract: Advanced atherosclerotic lesions contain senescent cells, but the role of these cells in atherogenesis remains unclear Using transgenic and pharmacological approaches to eliminate senescent cells in atherosclerosis-prone low-density lipoprotein receptor-deficient (Ldlr-/-) mice, we show that these cells are detrimental throughout disease pathogenesis We find that foamy macrophages with senescence markers accumulate in the subendothelial space at the onset of atherosclerosis, where they drive pathology by increasing expression of key atherogenic and inflammatory cytokines and chemokines In advanced lesions, senescent cells promote features of plaque instability, including elastic fiber degradation and fibrous cap thinning, by heightening metalloprotease production Together, these results demonstrate that senescent cells are key drivers of atheroma formation and maturation and suggest that selective clearance of these cells by senolytic agents holds promise for the treatment of atherosclerosis

Abstract: Event-triggered consensus of multiagent systems (MASs) has attracted tremendous attention from both theoretical and practical perspectives due to the fact that it enables all agents eventually to reach an agreement upon a common quantity of interest while significantly alleviating utilization of communication and computation resources. This paper aims to provide an overview of recent advances in event-triggered consensus of MASs. First, a basic framework of multiagent event-triggered operational mechanisms is established. Second, representative results and methodologies reported in the literature are reviewed and some in-depth analysis is made on several event-triggered schemes, including event-based sampling schemes, model-based event-triggered schemes, sampled-data-based event-triggered schemes, and self-triggered sampling schemes. Third, two examples are outlined to show applicability of event-triggered consensus in power sharing of microgrids and formation control of multirobot systems, respectively. Finally, some challenging issues on event-triggered consensus are proposed for future research.

Abstract: Material research has been a major driving force in the development of modern nano-electronic devices. In particular, research in magnetic thin films has revolutionized the development of spintronic devices; identifying new magnetic materials is key to better device performance and new device paradigm. The advent of two-dimensional van der Waals crystals creates new possibilities. This family of materials retain their chemical stability and structural integrity down to monolayers and, being atomically thin, are readily tuned by various kinds of gate modulation. Recent experiments have demonstrated that it is possible to obtain two-dimensional ferromagnetic order in insulating Cr$_2$Ge$_2$Te$_6$ and CrI$_3$ at low temperatures. Here, we developed a new device fabrication technique, and successfully isolated monolayers from layered metallic magnet Fe$_3$GeTe$_2$ for magnetotransport study. We found that the itinerant ferromagnetism persists in Fe$_3$GeTe$_2$ down to monolayer with an out-of-plane magnetocrystalline anisotropy. The ferromagnetic transition temperature, $T_c$, is suppressed in pristine Fe$_3$GeTe$_2$ thin flakes. An ionic gate, however, dramatically raises the $T_c$ up to room temperature, significantly higher than the bulk $T_c$ of 205 Kelvin. The gate-tunable room-temperature ferromagnetism in two-dimensional Fe$_3$GeTe$_2$ opens up opportunities for potential voltage-controlled magnetoelectronics based on atomically thin van der Waals crystals.

Abstract: Deep neural networks (DNNs) are one of the most prominent technologies of our time, as they achieve state-of-the-art performance in many machine learning tasks, including but not limited to image classification, text mining, and speech processing. However, recent research on DNNs has indicated ever-increasing concern on the robustness to adversarial examples, especially for security-critical tasks such as traffic sign identification for autonomous driving. Studies have unveiled the vulnerability of a well-trained DNN by demonstrating the ability of generating barely noticeable (to both human and machines) adversarial images that lead to misclassification. Furthermore, researchers have shown that these adversarial images are highly transferable by simply training and attacking a substitute model built upon the target model, known as a black-box attack to DNNs. Similar to the setting of training substitute models, in this paper we propose an effective black-box attack that also only has access to the input (images) and the output (confidence scores) of a targeted DNN. However, different from leveraging attack transferability from substitute models, we propose zeroth order optimization (ZOO) based attacks to directly estimate the gradients of the targeted DNN for generating adversarial examples. We use zeroth order stochastic coordinate descent along with dimension reduction, hierarchical attack and importance sampling techniques to efficiently attack black-box models. By exploiting zeroth order optimization, improved attacks to the targeted DNN can be accomplished, sparing the need for training substitute models and avoiding the loss in attack transferability. Experimental results on MNIST, CIFAR10 and ImageNet show that the proposed ZOO attack is as effective as the state-of-the-art white-box attack and significantly outperforms existing black-box attacks via substitute models.