Abstract: Having accurate, detailed, and up-to-date information about the location and behavior of animals in the wild would improve our ability to study and conserve ecosystems. We investigate the ability to automatically, accurately, and inexpensively collect such data, which could help catalyze the transformation of many fields of ecology, wildlife biology, zoology, conservation biology, and animal behavior into “big data” sciences. Motion-sensor “camera traps” enable collecting wildlife pictures inexpensively, unobtrusively, and frequently. However, extracting information from these pictures remains an expensive, time-consuming, manual task. We demonstrate that such information can be automatically extracted by deep learning, a cutting-edge type of artificial intelligence. We train deep convolutional neural networks to identify, count, and describe the behaviors of 48 species in the 3.2 million-image Snapshot Serengeti dataset. Our deep neural networks automatically identify animals with >93.8% accuracy, and we expect that number to improve rapidly in years to come. More importantly, if our system classifies only images it is confident about, our system can automate animal identification for 99.3% of the data while still performing at the same 96.6% accuracy as that of crowdsourced teams of human volunteers, saving >8.4 y (i.e., >17,000 h at 40 h/wk) of human labeling effort on this 3.2 million-image dataset. Those efficiency gains highlight the importance of using deep neural networks to automate data extraction from camera-trap images, reducing a roadblock for this widely used technology. Our results suggest that deep learning could enable the inexpensive, unobtrusive, high-volume, and even real-time collection of a wealth of information about vast numbers of animals in the wild.

Abstract: A warming climate is expected to have an impact on the magnitude and timing of river floods; however, no consistent large-scale climate change signal in observed flood magnitudes has been identified so far. We analyzed the timing of river floods in Europe over the past five decades, using a pan-European database from 4262 observational hydrometric stations, and found clear patterns of change in flood timing. Warmer temperatures have led to earlier spring snowmelt floods throughout northeastern Europe; delayed winter storms associated with polar warming have led to later winter floods around the North Sea and some sectors of the Mediterranean coast; and earlier soil moisture maxima have led to earlier winter floods in western Europe. Our results highlight the existence of a clear climate signal in flood observations at the continental scale.

Abstract: Convolutional networks have been the paradigm of choice in many computer vision applications. The convolution operation however has a significant weakness in that it only operates on a local neighborhood, thus missing global information. Self-attention, on the other hand, has emerged as a recent advance to capture long range interactions, but has mostly been applied to sequence modeling and generative modeling tasks. In this paper, we consider the use of self-attention for discriminative visual tasks as an alternative to convolutions. We introduce a novel two-dimensional relative self-attention mechanism that proves competitive in replacing convolutions as a stand-alone computational primitive for image classification. We find in control experiments that the best results are obtained when combining both convolutions and self-attention. We therefore propose to augment convolutional operators with this self-attention mechanism by concatenating convolutional feature maps with a set of feature maps produced via self-attention. Extensive experiments show that Attention Augmentation leads to consistent improvements in image classification on ImageNet and object detection on COCO across many different models and scales, including ResNets and a state-of-the art mobile constrained network, while keeping the number of parameters similar. In particular, our method achieves a $1.3\%$ top-1 accuracy improvement on ImageNet classification over a ResNet50 baseline and outperforms other attention mechanisms for images such as Squeeze-and-Excitation. It also achieves an improvement of 1.4 mAP in COCO Object Detection on top of a RetinaNet baseline.

Abstract: In this paper, we question if self-supervised learning provides new properties to Vision Transformer (ViT) that stand out compared to convolutional networks (convnets). Beyond the fact that adapting self-supervised methods to this architecture works particularly well, we make the following observations: first, self-supervised ViT features contain explicit information about the semantic segmentation of an image, which does not emerge as clearly with supervised ViTs, nor with convnets. Second, these features are also excellent k-NN classifiers, reaching 78.3% top-1 on ImageNet with a small ViT. Our study also underlines the importance of momentum encoder, multi-crop training, and the use of small patches with ViTs. We implement our findings into a simple self-supervised method, called DINO, which we interpret as a form of self-distillation with no labels. We show the synergy between DINO and ViTs by achieving 80.1% top-1 on ImageNet in linear evaluation with ViT-Base.

Abstract: We propose a novel direct sparse visual odometry formulation. It combines a fully direct probabilistic model (minimizing a photometric error) with consistent, joint optimization of all model parameters, including geometry -- represented as inverse depth in a reference frame -- and camera motion. This is achieved in real time by omitting the smoothness prior used in other direct methods and instead sampling pixels evenly throughout the images. Since our method does not depend on keypoint detectors or descriptors, it can naturally sample pixels from across all image regions that have intensity gradient, including edges or smooth intensity variations on mostly white walls. The proposed model integrates a full photometric calibration, accounting for exposure time, lens vignetting, and non-linear response functions. We thoroughly evaluate our method on three different datasets comprising several hours of video. The experiments show that the presented approach significantly outperforms state-of-the-art direct and indirect methods in a variety of real-world settings, both in terms of tracking accuracy and robustness.

Abstract: This paper surveys the recent attempts, both from the machine learning and operations research communities, at leveraging machine learning to solve combinatorial optimization problems. Given the hard nature of these problems, state-of-the-art algorithms rely on handcrafted heuristics for making decisions that are otherwise too expensive to compute or mathematically not well defined. Thus, machine learning looks like a natural candidate to make such decisions in a more principled and optimized way. We advocate for pushing further the integration of machine learning and combinatorial optimization and detail a methodology to do so. A main point of the paper is seeing generic optimization problems as data points and inquiring what is the relevant distribution of problems to use for learning on a given task.

Abstract: Discriminative Correlation Filters (DCF) are efficient in visual tracking but suffer from unwanted boundary effects. Spatially Regularized DCF (SRDCF) has been suggested to resolve this issue by enforcing spatial penalty on DCF coefficients, which, inevitably, improves the tracking performance at the price of increasing complexity. To tackle online updating, SRDCF formulates its model on multiple training images, further adding difficulties in improving efficiency. In this work, by introducing temporal regularization to SRDCF with single sample, we present our spatial-temporal regularized correlation filters (STRCF). The STRCF formulation can not only serve as a reasonable approximation to SRDCF with multiple training samples, but also provide a more robust appearance model than SRDCF in the case of large appearance variations. Besides, it can be efficiently solved via the alternating direction method of multipliers (ADMM). By incorporating both temporal and spatial regularization, our STRCF can handle boundary effects without much loss in efficiency and achieve superior performance over SRDCF in terms of accuracy and speed. Compared with SRDCF, STRCF with hand-crafted features provides a 5A— speedup and achieves a gain of 5.4% and 3.6% AUC score on OTB-2015 and Temple-Color, respectively. Moreover, STRCF with deep features also performs favorably against state-of-the-art trackers and achieves an AUC score of 68.3% on OTB-2015.

Abstract: BACKGROUND: For virtually every patient with colorectal cancer (CRC), hematoxylin-eosin (HE)-stained tissue slides are available. These images contain quantitative information, which is not routinely used to objectively extract prognostic biomarkers. In the present study, we investigated whether deep convolutional neural networks (CNNs) can extract prognosticators directly from these widely available images. METHODS AND FINDINGS: We hand-delineated single-tissue regions in 86 CRC tissue slides, yielding more than 100,000 HE image patches, and used these to train a CNN by transfer learning, reaching a nine-class accuracy of >94% in an independent data set of 7,180 images from 25 CRC patients. With this tool, we performed automated tissue decomposition of representative multitissue HE images from 862 HE slides in 500 stage I-IV CRC patients in the The Cancer Genome Atlas (TCGA) cohort, a large international multicenter collection of CRC tissue. Based on the output neuron activations in the CNN, we calculated a "deep stroma score," which was an independent prognostic factor for overall survival (OS) in a multivariable Cox proportional hazard model (hazard ratio [HR] with 95% confidence interval [CI]: 1.99 [1.27-3.12], p = 0.0028), while in the same cohort, manual quantification of stromal areas and a gene expression signature of cancer-associated fibroblasts (CAFs) were only prognostic in specific tumor stages. We validated these findings in an independent cohort of 409 stage I-IV CRC patients from the "Darmkrebs: Chancen der Verhutung durch Screening" (DACHS) study who were recruited between 2003 and 2007 in multiple institutions in Germany. Again, the score was an independent prognostic factor for OS (HR 1.63 [1.14-2.33], p = 0.008), CRC-specific OS (HR 2.29 [1.5-3.48], p = 0.0004), and relapse-free survival (RFS; HR 1.92 [1.34-2.76], p = 0.0004). A prospective validation is required before this biomarker can be implemented in clinical workflows. CONCLUSIONS: In our retrospective study, we show that a CNN can assess the human tumor microenvironment and predict prognosis directly from histopathological images.

Abstract: Hexagonal boron nitride (h-BN) is a natural hyperbolic material, in which the dielectric constants are the same in the basal plane (e(t) ≡ e(x) = e(y)) but have opposite signs (e(t)e(z) < 0) in the normal plane (e(z)). Owing to this property, finite-thickness slabs of h-BN act as multimode waveguides for the propagation of hyperbolic phonon polaritons--collective modes that originate from the coupling between photons and electric dipoles in phonons. However, control of these hyperbolic phonon polaritons modes has remained challenging, mostly because their electrodynamic properties are dictated by the crystal lattice of h-BN. Here we show, by direct nano-infrared imaging, that these hyperbolic polaritons can be effectively modulated in a van der Waals heterostructure composed of monolayer graphene on h-BN. Tunability originates from the hybridization of surface plasmon polaritons in graphene with hyperbolic phonon polaritons in h-BN, so that the eigenmodes of the graphene/h-BN heterostructure are hyperbolic plasmon-phonon polaritons. The hyperbolic plasmon-phonon polaritons in graphene/h-BN suffer little from ohmic losses, making their propagation length 1.5-2.0 times greater than that of hyperbolic phonon polaritons in h-BN. The hyperbolic plasmon-phonon polaritons possess the combined virtues of surface plasmon polaritons in graphene and hyperbolic phonon polaritons in h-BN. Therefore, graphene/h-BN can be classified as an electromagnetic metamaterial as the resulting properties of these devices are not present in its constituent elements alone.

Abstract: An opportunity in a crisis?

Abstract: Circulation. 2020;142:4–6. DOI: 10.1161/CIRCULATIONAHA.120.047659 4 Naveed Sattar , MD Iain B. McInnes, MD John J.V. McMurray, MD T he coronavirus disease 2019 (COVID-19) pandemic has led to worldwide research efforts to identify people at greatest risk of developing critical illness and dying. Initial data pointed toward older individuals being particularly vulnerable, as well as those with diabetes mellitus or cardiovascular (including hypertension), respiratory, or kidney disease. These problems are often concentrated in certain racial groups (eg, African Americans and Asians), which also appear to be more prone to worse COVID-19 outcomes.1 Increasing numbers of reports have linked obesity to more severe COVID-19 illness and death.1–3 In a French study, the risk for invasive mechanical ventilation in patients with COVID-19 infection admitted to the intensive treatment unit was more than 7-fold higher for those with body mass index (BMI) >35 compared with BMI <25 kg/m2.2 Among individuals with COVID-19 who were <60 years of age in New York City, those with a BMI between 30 to 34 kg/m2 and >35 kg/m2 were 1.8 times and 3.6 times more likely to be admitted to critical care, respectively, than individuals with a BMI <30 kg/m2.3 We suggest obesity or excess ectopic fat deposition may be a unifying risk factor for severe COVID-19 infection, reducing protective cardiorespiratory reserve as well as potentiating the immune dysregulation that appears, at least in part, to mediate the progression to critical illness and organ failure in a proportion of patients with COVID-19 (Figure). Whether obesity is an independent risk factor for susceptibility to infection requires further research. From a cardiovascular perspective, trial and genetic evidence conclusively show that obesity (and excess fat mass) are causally related to hypertension, diabetes mellitus, coronary heart disease, stroke, atrial fibrillation, renal disease, and heart failure. Obesity potentiates multiple cardiovascular risk factors, the premature development of cardiovascular disease, and adverse cardiorenal outcomes. There is also a metabolic concern. In individuals with diabetes mellitus, or at high risk of diabetes mellitus, obesity and excess ectopic fat lead to impairment of insulin resistance and reduced β-cell function. Both the latter limit ability to evoke an appropriate metabolic response on immunologic challenge, leading some patients with diabetes mellitus to require substantial amounts of insulin during severe infections. Overall, the integrated regulation of metabolism required for the complex cellular interactions, and for effective host defense, is lost, leading to functional immunologic deficit. COVID-19 may also directly disrupt pancreatic β-cell function through an interaction with angiotensin-converting enzyme 2. Furthermore, obesity enhances thrombosis, which is relevant given the association between severe COVID-19 and prothrombotic disseminated intravascular coagulation and high rates of venous thromboembolism. Beyond cardiometabolic and thrombotic consequences, obesity has detrimental effects on lung function, diminishing forced expiratory volume and forced vital capacity (Figure). Higher relative fat mass is also linked to such adverse changes, perhaps relevant to emerging reports of greater critical illness from COVID-19 in © 2020 American Heart Association, Inc. ON MY MIND

Abstract: Lithium–sulfur (Li–S) batteries have attracted much attention in the field of electrochemical energy storage and conversion. As a vital part of the cathode electrode, the host materials of sulfur usually have a strong impact on the capacity, energy density, cycle life and Coulombic efficiency of Li–S batteries. With their unique physical and chemical properties, the rationally designed hollow nanostructures show conspicuous advantages as sulfur hosts, and have significantly improved the overall performance of Li–S cells. The scope of this review considers the unique structural advantages of hollow host materials for high-performance Li–S batteries, together with a summary of recent advances in the design and synthesis of various hollow micro-/nanostructures with controlled shapes, tailored shell structures and designed chemical compositions. Finally, we propose some emerging requirements of sulfur hosts which we hope will shed some light on the future development trend of hollow structures for advanced Li–S batteries.

Abstract: Author(s): Kim, H; Kim, JC; Bianchini, M; Seo, DH; Rodriguez-Garcia, J; Ceder, G | Abstract: The development of rechargeable batteries using K ions as charge carriers has recently attracted considerable attention in the search for cost-effective and large-scale energy storage systems. In light of this trend, various materials for positive and negative electrodes are proposed and evaluated for application in K-ion batteries. Here, a comprehensive review of ongoing materials research on nonaqueous K-ion batteries is offered. Information on the status of new materials discovery and insights to help understand the K-storage mechanisms are provided. In addition, strategies to enhance the electrochemical properties of K-ion batteries and computational approaches to better understand their thermodynamic properties are included. Finally, K-ion batteries are compared to competing Li and Na systems and pragmatic opportunities and future research directions are discussed.

Abstract: Alzheimer’s disease (AD) is characterized by two molecular pathologies: cerebral β-amyloidosis in the form of β-amyloid (Aβ) plaques and tauopathy in the form of neurofibrillary tangles, neuritic plaques, and neuropil threads. Until recently, only Aβ could be studied in humans using positron emission tomography (PET) imaging owing to a lack of tau PET imaging agents. Clinical pathological studies have linked tau pathology closely to the onset and progression of cognitive symptoms in patients with AD. We report PET imaging of tau and Aβ in a cohort of cognitively normal older adults and those with mild AD. Multivariate analyses identified unique disease-related stereotypical spatial patterns (topographies) for deposition of tau and Aβ. These PET imaging tau and Aβ topographies were spatially distinct but correlated with disease progression. Cerebrospinal fluid measures of tau, often used to stage preclinical AD, correlated with tau deposition in the temporal lobe. Tau deposition in the temporal lobe more closely tracked dementia status and was a better predictor of cognitive performance than Aβ deposition in any region of the brain. These data support models of AD where tau pathology closely tracks changes in brain function that are responsible for the onset of early symptoms in AD.

Abstract: Metals, such as copper and silver, can be extremely toxic to bacteria at exceptionally low concentrations. Because of this biocidal activity, metals have been widely used as antimicrobial agents in a multitude of applications related with agriculture, healthcare, and the industry in general. Unlike other antimicrobial agents, metals are stable under conditions currently found in the industry allowing their use as additives. Today these metal based additives are found as: particles, ions absorbed/exchanged in different carriers, salts, hybrid structures, etc. One recent route to further extend the antimicrobial applications of these metals is by their incorporation as nanoparticles into polymer matrices. These polymer/metal nanocomposites can be prepared by several routes such as in situ synthesis of the nanoparticle within a hydrogel or direct addition of the metal nanofiller into a thermoplastic matrix. The objective of the present review is to show examples of polymer/metal composites designed to have antimicrobial activities, with a special focus on copper and silver metal nanoparticles and their mechanisms.

Abstract: The biological carbon pump is the process by which CO2 is transformed to organic carbon via photosynthesis, exported through sinking particles, and finally sequestered in the deep ocean. While the intensity of the pump correlates with plankton community composition, the underlying ecosystem structure driving the process remains largely uncharacterized. Here we use environmental and metagenomic data gathered during the Tara Oceans expedition to improve our understanding of carbon export in the oligotrophic ocean. We show that specific plankton communities, from the surface and deep chlorophyll maximum, correlate with carbon export at 150 m and highlight unexpected taxa such as Radiolaria and alveolate parasites, as well as Synechococcus and their phages, as lineages most strongly associated with carbon export in the subtropical, nutrient-depleted, oligotrophic ocean. Additionally, we show that the relative abundance of a few bacterial and viral genes can predict a significant fraction of the variability in carbon export in these regions.

Abstract: In this paper, we review some advances made recently in the study of mobile phone datasets. This area of research has emerged a decade ago, with the increasing availability of large-scale anonymized datasets, and has grown into a stand-alone topic. We survey the contributions made so far on the social networks that can be constructed with such data, the study of personal mobility, geographical partitioning, urban planning, and help towards development as well as security and privacy issues.

Abstract: Conversational speech recognition has served as a flagship speech recognition task since the release of the Switchboard corpus in the 1990s. In this paper, we measure the human error rate on the widely used NIST 2000 test set, and find that our latest automated system has reached human parity. The error rate of professional transcribers is 5.9% for the Switchboard portion of the data, in which newly acquainted pairs of people discuss an assigned topic, and 11.3% for the CallHome portion where friends and family members have open-ended conversations. In both cases, our automated system establishes a new state of the art, and edges past the human benchmark, achieving error rates of 5.8% and 11.0%, respectively. The key to our system's performance is the use of various convolutional and LSTM acoustic model architectures, combined with a novel spatial smoothing method and lattice-free MMI acoustic training, multiple recurrent neural network language modeling approaches, and a systematic use of system combination.

Abstract: Inside eukaryotic cells, macromolecules are partitioned into membrane-bounded compartments and, within these, some are further organized into non-membrane-bounded structures termed membrane-less organelles. The latter structures are comprised of heterogeneous mixtures of proteins and nucleic acids and assemble through a phase separation phenomenon similar to polymer condensation. Membrane-less organelles are dynamic structures maintained through multivalent interactions that mediate diverse biological processes, many involved in RNA metabolism. They rapidly exchange components with the cellular milieu and their properties are readily altered in response to environmental cues, often implicating membrane-less organelles in responses to stress signaling. In this review, we discuss: (1) the functional roles of membrane-less organelles, (2) unifying structural and mechanistic principles that underlie their assembly and disassembly, and (3) established and emerging methods used in structural investigations of membrane-less organelles.

Abstract: Most existing person re-identification (re-id) methods focus on learning the optimal distance metrics across camera views. Typically a person's appearance is represented using features of thousands of dimensions, whilst only hundreds of training samples are available due to the difficulties in collecting matched training images. With the number of training samples much smaller than the feature dimension, the existing methods thus face the classic small sample size (SSS) problem and have to resort to dimensionality reduction techniques and/or matrix regularisation, which lead to loss of discriminative power. In this work, we propose to overcome the SSS problem in re-id distance metric learning by matching people in a discriminative null space of the training data. In this null space, images of the same person are collapsed into a single point thus minimising the within-class scatter to the extreme and maximising the relative between-class separation simultaneously. Importantly, it has a fixed dimension, a closed-form solution and is very efficient to compute. Extensive experiments carried out on five person re-identification benchmarks including VIPeR, PRID2011, CUHK01, CUHK03 and Market1501 show that such a simple approach beats the state-of-the-art alternatives, often by a big margin.

Abstract: We report on the implementation of a tool for the analysis of second-order elastic stiffness tensors, provided with both an open-source Python module and a standalone online application allowing the visualization of anisotropic mechanical properties. After describing the software features, how we compute the conventional elastic constants and how we represent them graphically, we explain our technical choices for the implementation. In particular, we focus on why a Python module is used to generate the HTML web page with embedded Javascript for dynamical plots.

Abstract: Emerging Roles of Cell-free Tumor DNA in Oncology The capacity to detect new cancers, treatment-resistant variants, and tumor heterogeneity by noninvasive technology on the basis of tumor DNA in th...

Abstract: Reconstructing the phylogenetic relationships that unite all lineages (the tree of life) is a grand challenge. The paucity of homologous character data across disparately related lineages currently renders direct phylogenetic inference untenable. To reconstruct a comprehensive tree of life, we therefore synthesized published phylogenies, together with taxonomic classifications for taxa never incorporated into a phylogeny. We present a draft tree containing 2.3 million tips-the Open Tree of Life. Realization of this tree required the assembly of two additional community resources: (i) a comprehensive global reference taxonomy and (ii) a database of published phylogenetic trees mapped to this taxonomy. Our open source framework facilitates community comment and contribution, enabling the tree to be continuously updated when new phylogenetic and taxonomic data become digitally available. Although data coverage and phylogenetic conflict across the Open Tree of Life illuminate gaps in both the underlying data available for phylogenetic reconstruction and the publication of trees as digital objects, the tree provides a compelling starting point for community contribution. This comprehensive tree will fuel fundamental research on the nature of biological diversity, ultimately providing up-to-date phylogenies for downstream applications in comparative biology, ecology, conservation biology, climate change, agriculture, and genomics.

Abstract: Traffic waves are phenomena that emerge when the vehicular density exceeds a critical threshold. Considering the presence of increasingly automated vehicles in the traffic stream, a number of research activities have focused on the influence of automated vehicles on the bulk traffic flow. In the present article, we demonstrate experimentally that intelligent control of an autonomous vehicle is able to dampen stop-and-go waves that can arise even in the absence of geometric or lane changing triggers. Precisely, our experiments on a circular track with more than 20 vehicles show that traffic waves emerge consistently, and that they can be dampened by controlling the velocity of a single vehicle in the flow. We compare metrics for velocity, braking events, and fuel economy across experiments. These experimental findings suggest a paradigm shift in traffic management: flow control will be possible via a few mobile actuators (less than 5%) long before a majority of vehicles have autonomous capabilities.

Abstract: Larry Diamond is founding coeditor of the Journal of Democracy, se- nior fellow at the Hoover Institution and the Freeman Spogli Institute for International Studies at Stanford University, and director of Stan- ford's Center on Democracy, Development, and the Rule of Law. The year 2014 marked the fortieth anniversary of Portugal's Revolu- tion of the Carnations, which inaugurated what Samuel P. Huntington dubbed the "third wave" of global democratization. Any assessment of the state of global democracy today must begin by recognizing—even marveling at—the durability of this historic transformation. When the third wave began in 1974, only about 30 percent of the world's indepen- dent states met the criteria of electoral democracy—a system in which citizens, through universal suffrage, can choose and replace their leaders in regular, free, fair, and meaningful elections. 1 At that time, there were only about 46 democracies in the world. Most of those were the liberal democracies of the rich West, along with a number of small island states that had been British colonies. Only a few other developing democracies existed—principally, India, Sri Lanka, Costa Rica, Colombia, Venezu- ela, Israel, and Turkey. In the subsequent three decades, democracy had a remarkable global run, as the number of democracies essentially held steady or expanded every year from 1975 until 2007. Nothing like this continous growth in democracy had ever been seen before in the history of the world. While a number of these new "democracies" were quite illiberal—in some cases, so much so that Steven Levitsky and Lucan Way regard them as "competitive authoritarian" regimes 2 —the positive three-decade trend was paralleled by a similarly steady and significant expansion in levels of freedom (political rights and civil liberties, as measured annually by Freedom House). In 1974, the average level of freedom in the world stood at 4.38 (on the two seven-point scales, where 1 is most free and 7 is most repressive). It then gradually improved during the 1970s and