Showing papers by "Rensselaer Polytechnic Institute published in 2020"

Excess electronic recoil events in XENON1T

Abstract: We report results from searches for new physics with low-energy electronic recoil data recorded with the XENON1T detector. With an exposure of 0.65 tonne-years and an unprecedentedly low background rate of 76±2stat events/(tonne×year×keV) between 1 and 30 keV, the data enable one of the most sensitive searches for solar axions, an enhanced neutrino magnetic moment using solar neutrinos, and bosonic dark matter. An excess over known backgrounds is observed at low energies and most prominent between 2 and 3 keV. The solar axion model has a 3.4σ significance, and a three-dimensional 90% confidence surface is reported for axion couplings to electrons, photons, and nucleons. This surface is inscribed in the cuboid defined by gae<3.8×10-12, gaeganeff<4.8×10-18, and gaegaγ<7.7×10-22 GeV-1, and excludes either gae=0 or gaegaγ=gaeganeff=0. The neutrino magnetic moment signal is similarly favored over background at 3.2σ, and a confidence interval of μν∈(1.4,2.9)×10-11 μB (90% C.L.) is reported. Both results are in strong tension with stellar constraints. The excess can also be explained by β decays of tritium at 3.2σ significance with a corresponding tritium concentration in xenon of (6.2±2.0)×10-25 mol/mol. Such a trace amount can neither be confirmed nor excluded with current knowledge of its production and reduction mechanisms. The significances of the solar axion and neutrino magnetic moment hypotheses are decreased to 2.0σ and 0.9σ, respectively, if an unconstrained tritium component is included in the fitting. With respect to bosonic dark matter, the excess favors a monoenergetic peak at (2.3±0.2) keV (68% C.L.) with a 3.0σ global (4.0σ local) significance over background. This analysis sets the most restrictive direct constraints to date on pseudoscalar and vector bosonic dark matter for most masses between 1 and 210 keV/c2. We also consider the possibility that Ar37 may be present in the detector, yielding a 2.82 keV peak from electron capture. Contrary to tritium, the Ar37 concentration can be tightly constrained and is found to be negligible.

Macromolecular modeling and design in Rosetta: recent methods and frameworks

Abstract: The Rosetta software for macromolecular modeling, docking and design is extensively used in laboratories worldwide. During two decades of development by a community of laboratories at more than 60 institutions, Rosetta has been continuously refactored and extended. Its advantages are its performance and interoperability between broad modeling capabilities. Here we review tools developed in the last 5 years, including over 80 methods. We discuss improvements to the score function, user interfaces and usability. Rosetta is available at http://www.rosettacommons.org.

Deep learning in medical image registration: a survey

Abstract: The establishment of image correspondence through robust image registration is critical to many clinical tasks such as image fusion, organ atlas creation, and tumor growth monitoring and is a very challenging problem. Since the beginning of the recent deep learning renaissance, the medical imaging research community has developed deep learning-based approaches and achieved the state-of-the-art in many applications, including image registration. The rapid adoption of deep learning for image registration applications over the past few years necessitates a comprehensive summary and outlook, which is the main scope of this survey. This requires placing a focus on the different research areas as well as highlighting challenges that practitioners face. This survey, therefore, outlines the evolution of deep learning-based medical image registration in the context of both research challenges and relevant innovations in the past few years. Further, this survey highlights future research directions to show how this field may be possibly moved forward to the next level.

A Joint Neural Model for Information Extraction with Global Features.

Abstract: Most existing joint neural models for Information Extraction (IE) use local task-specific classifiers to predict labels for individual instances (e.g., trigger, relation) regardless of their interactions. For example, a victim of a die event is likely to be a victim of an attack event in the same sentence. In order to capture such cross-subtask and cross-instance inter-dependencies, we propose a joint neural framework, OneIE, that aims to extract the globally optimal IE result as a graph from an input sentence. OneIE performs end-to-end IE in four stages: (1) Encoding a given sentence as contextualized word representations; (2) Identifying entity mentions and event triggers as nodes; (3) Computing label scores for all nodes and their pairwise links using local classifiers; (4) Searching for the globally optimal graph with a beam decoder. At the decoding stage, we incorporate global features to capture the cross-subtask and cross-instance interactions. Experiments show that adding global features improves the performance of our model and achieves new state of-the-art on all subtasks. In addition, as OneIE does not use any language-specific feature, we prove it can be easily applied to new languages or trained in a multilingual manner.

CT Super-Resolution GAN Constrained by the Identical, Residual, and Cycle Learning Ensemble (GAN-CIRCLE)

Abstract: In this paper, we present a semi-supervised deep learning approach to accurately recover high-resolution (HR) CT images from low-resolution (LR) counterparts. Specifically, with the generative adversarial network (GAN) as the building block, we enforce the cycle-consistency in terms of the Wasserstein distance to establish a nonlinear end-to-end mapping from noisy LR input images to denoised and deblurred HR outputs. We also include the joint constraints in the loss function to facilitate structural preservation. In this process, we incorporate deep convolutional neural network (CNN), residual learning, and network in network techniques for feature extraction and restoration. In contrast to the current trend of increasing network depth and complexity to boost the imaging performance, we apply a parallel ${1}\times {1}$ CNN to compress the output of the hidden layer and optimize the number of layers and the number of filters for each convolutional layer. The quantitative and qualitative evaluative results demonstrate that our proposed model is accurate, efficient and robust for super-resolution (SR) image restoration from noisy LR input images. In particular, we validate our composite SR networks on three large-scale CT datasets, and obtain promising results as compared to the other state-of-the-art methods.

Deep learning for tomographic image reconstruction

Abstract: Deep-learning-based tomographic imaging is an important application of artificial intelligence and a new frontier of machine learning. Deep learning has been widely used in computer vision and image analysis, which deal with existing images, improve these images, and produce features from them. Since 2016, deep learning techniques have been actively researched for tomographic imaging, especially in the context of biomedicine, with impressive results and great potential. Tomographic reconstruction produces images of multi-dimensional structures from externally measured ‘encoded’ data in the form of various tomographic transforms (integrals, harmonics, echoes and so on). In this Review, we provide a general background, highlight representative results with an emphasis on medical imaging, and discuss key issues that need to be addressed in this emerging field. In particular, tomographic imaging is an integral part of modern medicine, and will play a key role in personalized, preventive and precision medicine and make it intelligent, inexpensive and indiscriminate. The popularity of deep learning is leading to new areas in biomedical applications. Wang and colleagues summarize in this Review the recent development and future directions of deep neural networks for superior image quality in the tomographic imaging field.

The CARE Principles for Indigenous Data Governance

Abstract: Concerns about secondary use of data and limited opportunities for benefit-sharing have focused attention on the tension that Indigenous communities feel between (1) protecting Indigenous rights and interests in Indigenous data (including traditional knowledges) and (2) supporting open data, machine learning, broad data sharing, and big data initiatives. The International Indigenous Data Sovereignty Interest Group (within the Research Data Alliance) is a network of nation-state based Indigenous data sovereignty networks and individuals that developed the ‘CARE Principles for Indigenous Data Governance’ (Collective Benefit, Authority to Control, Responsibility, and Ethics) in consultation with Indigenous Peoples, scholars, non-profit organizations, and governments. The CARE Principles are people– and purpose-oriented, reflecting the crucial role of data in advancing innovation, governance, and self-determination among Indigenous Peoples. The Principles complement the existing data-centric approach represented in the ‘FAIR Guiding Principles for scientific data management and stewardship’ (Findable, Accessible, Interoperable, Reusable). The CARE Principles build upon earlier work by the Te Mana Raraunga Maori Data Sovereignty Network, US Indigenous Data Sovereignty Network, Maiam nayri Wingara Aboriginal and Torres Strait Islander Data Sovereignty Collective, and numerous Indigenous Peoples, nations, and communities. The goal is that stewards and other users of Indigenous data will ‘Be FAIR and CARE.’ In this first formal publication of the CARE Principles, we articulate their rationale, describe their relation to the FAIR Principles, and present examples of their application.

Sulfated polysaccharides effectively inhibit SARS-CoV-2 in vitro

Abstract: Dear Editor, COVID-19, caused by the SARS-CoV-2 virus, has now spread worldwide with catastrophic human and economic impacts and currently has infected over 10 million people and killed over 500,000. In an effort to mitigate disease symptoms and impede viral spread, efforts in vaccine development and drug discovery are being conducted at a rapid pace. Recently, we showed that the well-known anticoagulant heparin has exceptional binding affinity to the spike protein (S-protein) of SARS-CoV-2. The Sprotein of SARS-CoV-2 bound more tightly to immobilized heparin (KD= ~10 −11 M) than the S-proteins of either SARS-CoV (KD= ~10 −7 M) or MERS-CoV (KD= ~10 M). However, it is not known whether the tight binding of heparin to the SARS-CoV-2 S-protein translates into potent antiviral activity. In the current study, we evaluated the in vitro antiviral properties of heparin and other closely related polysaccharides to assess the relevance of heparin-related GAGs and other sulfated polysaccharides as part of the pharmacopeia of potential therapeutics that target SARS-CoV-2. Vero-CCL81, which expresses both ACE2 and TMPRSS2, were used for viral replication at high titer for use in antiviral assays. Heparin, heparan sulfates, other glycosaminoglycans (GAGs), and fucoidan and other highly sulfated polysaccharides were screened using surface plasmon resonance (SPR) to measure binding affinity to the SARSCoV-2 S-protein (Fig. 1a). Briefly, solution competition studies between surface immobilized heparin and other sulfated polysaccharides were evaluated by injecting SARS-CoV-2 S-protein (50 nM) alone or mixed with 1 μM of an indicated polysaccharide in SPR buffer at a flow rate of 30 μL/min. After each run, dissociation and regeneration were performed. For each set of competition experiments, a control experiment (S-protein without polysaccharide) was performed to ensure the surface was fully regenerated. Among the tested polysaccharides, RPI27 and RPI-28, complex sulfated polysaccharides (fucoidans) extracted from the seaweed Saccharina japonica, chemo-enzymatically synthesized trisulfated (TriS) heparin, and unfractionated USP-heparin itself were able to compete with heparin for S-protein binding. We selected these compounds along with a non-anticoagulant low molecular weight heparin (NACH) for further study (Fig. 1b). The other GAGs including heparan sulfate, the chondroitin sulfates, and keratan sulfate show no competitive binding when compared to the control. Standard assays were performed to quantify potential cytotoxicity and antiviral activity. Cytotoxicity determination of the polysaccharides was performed using Vero cells and the standard water-soluble tetrazolium salt-1 (WST-1) assay (Takara Bio Inc., Japan). None of the tested polysaccharides showed toxicity even at the highest concentrations tested. Vero cells were infected with SARS-CoV-2 at a multiplicity of infection (MOI) of 2.5 × 10 with varying dosages of polysaccharide to confirm antiviral activity. A focus reduction assay was performed 48 h post infection to determine efficacy. Antiviral activities correlated with the SPR results. The most potent compound tested, RPI-27, is a high molecular weight, branched polysaccharide related to the known compound fucoidan, and had an EC50 of 8.3 ± 4.6 μg/mL, which corresponds to ~83 nM (Fig. 1c, d and Supplementary Table S1). This is substantially more potent than

Projected WIMP sensitivity of the XENONnT dark matter experiment

Abstract: XENONnT is a dark matter direct detection experiment, utilizing 5.9 t of instrumented liquid xenon, located at the INFN Laboratori Nazionali del Gran Sasso. In this work, we predict the experimental background and project the sensitivity of XENONnT to the detection of weakly interacting massive particles (WIMPs). The expected average differential background rate in the energy region of interest, corresponding to (1, 13) keV and (4, 50) keV for electronic and nuclear recoils, amounts to 12.3 ± 0.6 (keV t y)-1 and (2.2± 0.5)× 10−3 (keV t y)-1, respectively, in a 4 t fiducial mass. We compute unified confidence intervals using the profile construction method, in order to ensure proper coverage. With the exposure goal of 20 t y, the expected sensitivity to spin-independent WIMP-nucleon interactions reaches a cross-section of 1.4×10−48 cm2 for a 50 GeV/c2 mass WIMP at 90% confidence level, more than one order of magnitude beyond the current best limit, set by XENON1T . In addition, we show that for a 50 GeV/c2 WIMP with cross-sections above 2.6×10−48 cm2 (5.0×10−48 cm2) the median XENONnT discovery significance exceeds 3σ (5σ). The expected sensitivity to the spin-dependent WIMP coupling to neutrons (protons) reaches 2.2×10−43 cm2 (6.0×10−42 cm2).

Porous Two-Dimensional Materials for Photocatalytic and Electrocatalytic Applications

Abstract: Summary Two-dimensional materials with abundant in-plane pores (porous 2D materials) have shown high performances as catalysts, especially for photocatalysis and electrocatalysis, owing to their distinct microstructural advantages originating from both 2D materials and porous materials. Here, the recent progress in porous 2D materials in photocatalysis and electrocatalysis is reviewed. We first highlight the influence of their special structural merits on the processes of photocatalysis and electrocatalysis, including transport of ion and/or charge carriers, surface active sites, stability, modifications, electronic band structure, and light absorption properties. Representative synthetic methods for porous 2D materials are also introduced classified by top-down and bottom-up routes. In addition, their applications in different aspects of photocatalysis and electrocatalysis are presented systematically. In conclusion, we propose some opportunities and challenges for the development of porous 2D materials, with the hope of further facilitating the applications of these emerging advanced materials in photocatalysis and electrolysis.

ABINIT: Overview and focus on selected capabilities.

Abstract: abinit is probably the first electronic-structure package to have been released under an open-source license about 20 years ago. It implements density functional theory, density-functional perturbation theory (DFPT), many-body perturbation theory (GW approximation and Bethe-Salpeter equation), and more specific or advanced formalisms, such as dynamical mean-field theory (DMFT) and the "temperature-dependent effective potential" approach for anharmonic effects. Relying on planewaves for the representation of wavefunctions, density, and other space-dependent quantities, with pseudopotentials or projector-augmented waves (PAWs), it is well suited for the study of periodic materials, although nanostructures and molecules can be treated with the supercell technique. The present article starts with a brief description of the project, a summary of the theories upon which abinit relies, and a list of the associated capabilities. It then focuses on selected capabilities that might not be present in the majority of electronic structure packages either among planewave codes or, in general, treatment of strongly correlated materials using DMFT; materials under finite electric fields; properties at nuclei (electric field gradient, Mossbauer shifts, and orbital magnetization); positron annihilation; Raman intensities and electro-optic effect; and DFPT calculations of response to strain perturbation (elastic constants and piezoelectricity), spatial dispersion (flexoelectricity), electronic mobility, temperature dependence of the gap, and spin-magnetic-field perturbation. The abinit DFPT implementation is very general, including systems with van der Waals interaction or with noncollinear magnetism. Community projects are also described: generation of pseudopotential and PAW datasets, high-throughput calculations (databases of phonon band structure, second-harmonic generation, and GW computations of bandgaps), and the library libpaw. abinit has strong links with many other software projects that are briefly mentioned.

International scientists formulate a roadmap for insect conservation and recovery

Abstract: To the Editor — A growing number of studies are providing evidence that a suite of anthropogenic stressors — habitat loss and fragmentation, pollution, invasive species, climate change and overharvesting — are seriously reducing insect and other invertebrate abundance, diversity and biomass across the biosphere1–8. These declines affect all functional groups: herbivores, detritivores, parasitoids, predators and pollinators. Insects are vitally important in a wide range of ecosystem services9 of which some are vitally important for food production and security (for example, pollination and pest control)10. There is now a strong scientific consensus that the decline of insects, other arthropods and biodiversity as a whole, is a very real and serious threat that society must urgently address11–13. In response to the increasing public awareness of the problem, the German government is committing funds to combat and reverse declining insect numbers13. This funding should act as a clarion call to other nations across the world — especially wealthier ones — to follow suit and to respond proactively to the crisis by addressing the known and suspected threats and implementing solutions. We hereby propose a global ‘roadmap’ for insect conservation and recovery (Fig. 1). This entails the immediate implementation of several ‘no-regret’ measures (Fig. 1, step 1) that will act to slow or stop insect declines. Among the initiatives we encourage are the following immediate measures: Taking aggressive steps to reduce greenhouse gas emissions; reversing recent trends in agricultural intensification including reduced application of synthetic pesticides and fertilizers and pursuing their replacement with agro-ecological measures; promoting the diversification and maintenance of locally adapted landuse techniques; increasing landscape heterogeneity through the maintenance of natural areas within the landscape matrix and ensuring the retention and creation of microhabitats within habitats which may be increasingly important for insects during extreme climatic events such as droughts or heatwaves; reducing identified local threats such as light, water or noise pollution, invasive species and so on; prioritizing the import of goods that are not produced at the cost of healthy, species-rich ecosystems; designing and deploying policies (for example, subsidies and taxation) to induce the innovation and adoption of insectfriendly technologies; enforcing stricter measures to reduce the introduction of alien species, and prioritizing nature-based tactics for their (long-term) mitigation; compiling and implementing conservation strategies for species that are vulnerable, threatened or endangered; funding educational and outreach programs, including those tailored to the needs of the wider public, farmers, land managers, decision makers and conservation professionals; enhancing ‘citizen science’ or ‘community science’ as a way of obtaining more data on insect diversity and abundance as well as engaging the public, especially in areas where academic or professional infrastructure is lacking; devising and deploying measures across agricultural and food value chains that favour insect-friendly farming, including tracking, labelling, certification and insurance schemes or outcome-based incentives that facilitate behavioural changes, and investing in capacity building to create a new generation of insect conservationists and providing knowledge and skills to existing professionals (particularly in developing countries). To better understand changes in insect abundance and diversity, research should aim to prioritize the following areas: Quantifying temporal trends in insect abundance, diversity and biomass by extracting long-term datasets from existing insect collections to inform new censuses; exploring the relative contributions of different anthropogenic stressors causing insect declines within and across different taxa; initiating long-term studies comparing insect abundance and diversity in different habitats and ecosystems along a management-intensity gradient and at the intersection of agricultural and natural habitats; designing and validating insectfriendly techniques that are effective, locally relevant and economically sound in agriculture, managed habitats and urban environments; promoting and applying standardized monitoring protocols globally and establishing long-term monitoring plots or sites based on such protocols, as well as increasing support for existing monitoring efforts; establishing an international governing body under the auspices of existing bodies (for example, the United Nations Environment Programme (UNEP) or the International Union for Conservation of Nature (IUCN)) that is accountable for documenting and monitoring the effects of proposed solutions on insect biodiversity in the longer term; launching public–private partnerships and sustainable financing initiatives with the aim of restoring, protecting and creating new vital insect habitats as well as managing key threats; increasing exploration and research to improve biodiversity assessments, with a focus on regional capacity building in understudied and neglected areas, and performing large-scale assessments of the conservation status of insect groups to help define priority species, areas and issues. Most importantly, we should not wait to act until we have addressed every key knowledge gap. We currently have enough information on some key causes of insect decline to formulate no-regret solutions whilst more data are compiled for lesserknown taxa and regions and long-term data are aggregated and assessed. Implementation should be accompanied by research that examines impacts, the results of which can be used to modify and improve the implementation of effective measures. Furthermore, such a ‘learning-by-doing’ approach ensures that these conservation strategies are robust to newly emerging pressures and threats. We must act now. ❐

Iterative Deep Graph Learning for Graph Neural Networks: Better and Robust Node Embeddings

Abstract: In this paper, we propose an end-to-end graph learning framework, namely Iterative Deep Graph Learning (IDGL), for jointly and iteratively learning graph structure and graph embedding. The key rationale of IDGL is to learn a better graph structure based on better node embeddings, and vice versa (i.e., better node embeddings based on a better graph structure). Our iterative method dynamically stops when the learned graph structure approaches close enough to the graph optimized for the downstream prediction task. In addition, we cast the graph learning problem as a similarity metric learning problem and leverage adaptive graph regularization for controlling the quality of the learned graph. Finally, combining the anchor-based approximation technique, we further propose a scalable version of IDGL, namely IDGL-Anch, which significantly reduces the time and space complexity of IDGL without compromising the performance. Our extensive experiments on nine benchmarks show that our proposed IDGL models can consistently outperform or match the state-of-the-art baselines. Furthermore, IDGL can be more robust to adversarial graphs and cope with both transductive and inductive learning.

Designer DNA architecture offers precise and multivalent spatial pattern-recognition for viral sensing and inhibition.

Abstract: DNA, when folded into nanostructures with a specific shape, is capable of spacing and arranging binding sites into a complex geometric pattern with nanometre precision. Here we demonstrate a designer DNA nanostructure that can act as a template to display multiple binding motifs with precise spatial pattern-recognition properties, and that this approach can confer exceptional sensing and potent viral inhibitory capabilities. A star-shaped DNA architecture, carrying five molecular beacon-like motifs, was constructed to display ten dengue envelope protein domain III (ED3)-targeting aptamers into a two-dimensional pattern precisely matching the spatial arrangement of ED3 clusters on the dengue (DENV) viral surface. The resulting multivalent interactions provide high DENV-binding avidity. We show that this structure is a potent viral inhibitor and that it can act as a sensor by including a fluorescent output to report binding. Our molecular-platform design strategy could be adapted to detect and combat other disease-causing pathogens by generating the requisite ligand patterns on customized DNA nanoarchitectures. DNA is capable of self-assembling into a wide range of user-defined structures and so can be used as a scaffold to arrange binding motifs with nanometre precision. Now, DNA has been used to accurately display aptamers that fit the repeated epitope pattern of a dengue viral antigen to produce a nanostructure that can be a potent viral inhibitor or a fluorescent sensor.

High Interfacial-Energy Interphase Promoting Safe Lithium Metal Batteries.

Abstract: Engineering a stable solid electrolyte interphase (SEI) is critical for suppression of lithium dendrites. However, the formation of a desired SEI by formulating electrolyte composition is very difficult due to complex electrochemical reduction reactions. Here, instead of trial-and-error of electrolyte composition, we design a Li-11 wt % Sr alloy anode to form a SrF2-rich SEI in fluorinated electrolytes. Density functional theory (DFT) calculation and experimental characterization demonstrate that a SrF2-rich SEI has a large interfacial energy with Li metal and a high mechanical strength, which can effectively suppress the Li dendrite growth by simultaneously promoting the lateral growth of deposited Li metal and the SEI stability. The Li-Sr/Cu cells in 2 M LiFSI-DME show an outstanding Li plating/stripping Coulombic efficiency of 99.42% at 1 mA cm-2 with a capacity of 1 mAh cm-2 and 98.95% at 3 mA cm-2 with a capacity of 2 mAh cm-2, respectively. The symmetric Li-Sr/Li-Sr cells also achieve a stable electrochemical performance of 180 cycles at an extremely high current density of 30 mA cm-2 with a capacity of 1 mAh cm-2. When paired with LiFePO4 (LFP) and LiNi0.8Co0.1Mn0.1O2 (NCM811) cathodes, Li-Sr/LFP cells in 2 M LiFSI-DME electrolytes and Li-Sr/NMC811 cells in 1 M LiPF6 in FEC:FEMC:HFE electrolytes also maintain excellent capacity retention. Designing SEIs by regulating Li-metal anode composition opens up a new and rational avenue to suppress Li dendrites.

Three Dimensional Bioprinting of a Vascularized and Perfusable Skin Graft Using Human Keratinocytes, Fibroblasts, Pericytes, and Endothelial Cells

Abstract: Multilayered skin substitutes comprising allogeneic cells have been tested for the treatment of nonhealing cutaneous ulcers. However, such nonnative skin grafts fail to permanently engraft because they lack dermal vascular networks important for integration with the host tissue. In this study, we describe the fabrication of an implantable multilayered vascularized bioengineered skin graft using 3D bioprinting. The graft is formed using one bioink containing human foreskin dermal fibroblasts (FBs), human endothelial cells (ECs) derived from cord blood human endothelial colony-forming cells (HECFCs), and human placental pericytes (PCs) suspended in rat tail type I collagen to form a dermis followed by printing with a second bioink containing human foreskin keratinocytes (KCs) to form an epidermis. In vitro, KCs replicate and mature to form a multilayered barrier, while the ECs and PCs self-assemble into interconnected microvascular networks. The PCs in the dermal bioink associate with EC-lined vascular structures and appear to improve KC maturation. When these 3D printed grafts are implanted on the dorsum of immunodeficient mice, the human EC-lined structures inosculate with mouse microvessels arising from the wound bed and become perfused within 4 weeks after implantation. The presence of PCs in the printed dermis enhances the invasion of the graft by host microvessels and the formation of an epidermal rete. Impact Statement Three Dimensional printing can be used to generate multilayered vascularized human skin grafts that can potentially overcome the limitations of graft survival observed in current avascular skin substitutes. Inclusion of human pericytes in the dermal bioink appears to improve both dermal and epidermal maturation.

The search for the most conductive metal for narrow interconnect lines

Abstract: A major challenge for the continued downscaling of integrated circuits is the resistivity increase of Cu interconnect lines with decreasing dimensions. Alternative metals have the potential to mitigate this resistivity bottleneck by either (a) facilitating specular electron interface scattering and negligible grain boundary reflection or (b) a low bulk mean free path that renders resistivity scaling negligible. Recent research suggests that specular electron scattering at the interface between the interconnect metal and the liner layer requires a low density of states at the interface and in the liner (i.e., an insulating liner) and either a smooth epitaxial metal-liner interface or only weak van der Waals bonding as typical for 2D liner materials. The grain boundary contribution to the room-temperature resistivity becomes negligible if the grain size is large (>200 nm or ten times the linewidth for wide or narrow conductors, respectively) or if the electron reflection coefficient is small due to low-energy boundaries and electronic state matching of neighboring grains. First-principles calculations provide a list of metals (Rh, Pt, Ir, Nb, Ru, Ni, etc.) with a small product of the bulk resistivity times the bulk electron mean free path ρo × λ, which is an indicator for suppressed resistivity scaling. However, resistivity measurements on epitaxial layers indicate considerably larger experimental ρo × λ values for many metals, indicating the breakdown of the classical transport models at small (<10 nm) dimensions and suggesting that Ir is the most promising elemental metal for narrow high-conductivity interconnects, followed by Ru and Rh.

Boundary-Weighted Domain Adaptive Neural Network for Prostate MR Image Segmentation

Abstract: Accurate segmentation of the prostate from magnetic resonance (MR) images provides useful information for prostate cancer diagnosis and treatment. However, automated prostate segmentation from 3D MR images faces several challenges. The lack of clear edge between the prostate and other anatomical structures makes it challenging to accurately extract the boundaries. The complex background texture and large variation in size, shape and intensity distribution of the prostate itself make segmentation even further complicated. Recently, as deep learning, especially convolutional neural networks (CNNs), emerging as the best performed methods for medical image segmentation, the difficulty in obtaining large number of annotated medical images for training CNNs has become much more pronounced than ever. Since large-scale dataset is one of the critical components for the success of deep learning, lack of sufficient training data makes it difficult to fully train complex CNNs. To tackle the above challenges, in this paper, we propose a boundary-weighted domain adaptive neural network (BOWDA-Net). To make the network more sensitive to the boundaries during segmentation, a boundary-weighted segmentation loss is proposed. Furthermore, an advanced boundary-weighted transfer leaning approach is introduced to address the problem of small medical imaging datasets. We evaluate our proposed model on three different MR prostate datasets. The experimental results demonstrate that the proposed model is more sensitive to object boundaries and outperformed other state-of-the-art methods.

Na+-gated water-conducting nanochannels for boosting CO2 conversion to liquid fuels

Abstract: Robust, gas-impeding water-conduction nanochannels that can sieve water from small gas molecules such as hydrogen (H2), particularly at high temperature and pressure, are desirable for boosting many important reactions severely restricted by water (the major by-product) both thermodynamically and kinetically. Identifying and constructing such nanochannels into large-area separation membranes without introducing extra defects is challenging. We found that sodium ion (Na+)–gated water-conduction nanochannels could be created by assembling NaA zeolite crystals into a continuous, defect-free separation membrane through a rationally designed method. Highly efficient in situ water removal through water-conduction nanochannels led to a substantial increase in carbon dioxide (CO2) conversion and methanol yield in CO2 hydrogenation for methanol production.

On Interpretability of Artificial Neural Networks: A Survey

Abstract: Deep learning as represented by the artificial deep neural networks (DNNs) has achieved great success in many important areas that deal with text, images, videos, graphs, and so on. However, the black-box nature of DNNs has become one of the primary obstacles for their wide acceptance in mission-critical applications such as medical diagnosis and therapy. Due to the huge potential of deep learning, interpreting neural networks has recently attracted much research attention. In this paper, based on our comprehensive taxonomy, we systematically review recent studies in understanding the mechanism of neural networks, describe applications of interpretability especially in medicine, and discuss future directions of interpretability research, such as in relation to fuzzy logic and brain science.

Towards Visually Explaining Variational Autoencoders

Abstract: Recent advances in Convolutional Neural Network (CNN) model interpretability have led to impressive progress in visualizing and understanding model predictions. In particular, gradient-based visual attention methods have driven much recent effort in using visual attention maps as a means for visual explanations. A key problem, however, is these methods are designed for classification and categorization tasks, and their extension to explaining generative models, e.g., variational autoencoders (VAE) is not trivial. In this work, we take a step towards bridging this crucial gap, proposing the first technique to visually explain VAEs by means of gradient-based attention. We present methods to generate visual attention from the learned latent space, and also demonstrate such attention explanations serve more than just explaining VAE predictions. We show how these attention maps can be used to localize anomalies in images, demonstrating state-of-the-art performance on the MVTec-AD dataset. We also show how they can be infused into model training, helping bootstrap the VAE into learning improved latent space disentanglement, demonstrated on the Dsprites dataset.

Single-atom doping of MoS2 with manganese enables ultrasensitive detection of dopamine: Experimental and computational approach.

Abstract: Two-dimensional transition metal dichalcogenides (TMDs) emerged as a promising platform to construct sensitive biosensors. We report an ultrasensitive electrochemical dopamine sensor based on manganese-doped MoS2 synthesized via a scalable two-step approach (with Mn ~2.15 atomic %). Selective dopamine detection is achieved with a detection limit of 50 pM in buffer solution, 5 nM in 10% serum, and 50 nM in artificial sweat. Density functional theory calculations and scanning transmission electron microscopy show that two types of Mn defects are dominant: Mn on top of a Mo atom (MntopMo) and Mn substituting a Mo atom (MnMo). At low dopamine concentrations, physisorption on MnMo dominates. At higher concentrations, dopamine chemisorbs on MntopMo, which is consistent with calculations of the dopamine binding energy (2.91 eV for MntopMo versus 0.65 eV for MnMo). Our results demonstrate that metal-doped layered materials, such as TMDs, constitute an emergent platform to construct ultrasensitive and tunable biosensors.

Multi-Organ Segmentation Over Partially Labeled Datasets With Multi-Scale Feature Abstraction

Abstract: Shortage of fully annotated datasets has been a limiting factor in developing deep learning based image segmentation algorithms and the problem becomes more pronounced in multi-organ segmentation. In this paper, we propose a unified training strategy that enables a novel multi-scale deep neural network to be trained on multiple partially labeled datasets for multi-organ segmentation. In addition, a new network architecture for multi-scale feature abstraction is proposed to integrate pyramid input and feature analysis into a U-shape pyramid structure. To bridge the semantic gap caused by directly merging features from different scales, an equal convolutional depth mechanism is introduced. Furthermore, we employ a deep supervision mechanism to refine the outputs in different scales. To fully leverage the segmentation features from all the scales, we design an adaptive weighting layer to fuse the outputs in an automatic fashion. All these mechanisms together are integrated into a Pyramid Input Pyramid Output Feature Abstraction Network (PIPO-FAN). Our proposed method was evaluated on four publicly available datasets, including BTCV, LiTS, KiTS and Spleen, where very promising performance has been achieved. The source code of this work is publicly shared at https://github.com/DIAL-RPI/PIPO-FAN to facilitate others to reproduce the work and build their own models using the introduced mechanisms.

Ultrafast hot-hole injection modifies hot-electron dynamics in Au/p-GaN heterostructures

Abstract: A fundamental understanding of hot-carrier dynamics in photo-excited metal nanostructures is needed to unlock their potential for photodetection and photocatalysis. Despite numerous studies on the ultrafast dynamics of hot electrons, so far, the temporal evolution of hot holes in metal–semiconductor heterostructures remains unknown. Here, we report ultrafast (t < 200 fs) hot-hole injection from Au nanoparticles into the valence band of p-type GaN. The removal of hot holes from below the Au Fermi level is observed to substantially alter the thermalization dynamics of hot electrons, reducing the peak electronic temperature and the electron–phonon coupling time of the Au nanoparticles. First-principles calculations reveal that hot-hole injection modifies the relaxation dynamics of hot electrons in Au nanoparticles by modulating the electronic structure of the metal on timescales commensurate with electron–electron scattering. These results advance our understanding of hot-hole dynamics in metal–semiconductor heterostructures and offer additional strategies for manipulating the dynamics of hot carriers on ultrafast timescales. Photo-excited gold nanoparticles are shown to provide ultrafast and efficient hot-hole injection to the valence band of p-type GaN, substantially altering hot-electron dynamics in the nanoparticles and forming a basis to design hot-hole-based optoelectronics.

Enabling room temperature ferromagnetism in monolayer MoS 2 via in situ iron-doping

Abstract: Two-dimensional semiconductors, including transition metal dichalcogenides, are of interest in electronics and photonics but remain nonmagnetic in their intrinsic form. Previous efforts to form two-dimensional dilute magnetic semiconductors utilized extrinsic doping techniques or bulk crystal growth, detrimentally affecting uniformity, scalability, or Curie temperature. Here, we demonstrate an in situ substitutional doping of Fe atoms into MoS2 monolayers in the chemical vapor deposition growth. The iron atoms substitute molybdenum sites in MoS2 crystals, as confirmed by transmission electron microscopy and Raman signatures. We uncover an Fe-related spectral transition of Fe:MoS2 monolayers that appears at 2.28 eV above the pristine bandgap and displays pronounced ferromagnetic hysteresis. The microscopic origin is further corroborated by density functional theory calculations of dipole-allowed transitions in Fe:MoS2. Using spatially integrating magnetization measurements and spatially resolving nitrogen-vacancy center magnetometry, we show that Fe:MoS2 monolayers remain magnetized even at ambient conditions, manifesting ferromagnetism at room temperature. Ferromagnetism with a Curie temperature above room temperature in 2D materials is highly desirable for practical spintronics applications. Here, the authors demonstrate such phenomenon in monolayer MoS2 via in situ iron-doping and measured local magnetic field strength up to 0.5 ± 0.1 mT.

Practical Detection of Trojan Neural Networks: Data-Limited and Data-Free Cases

Abstract: When the training data are maliciously tampered, the predictions of the acquired deep neural network (DNN) can be manipulated by an adversary known as the Trojan attack (or poisoning backdoor attack). The lack of robustness of DNNs against Trojan attacks could significantly harm real-life machine learning (ML) systems in downstream applications, therefore posing widespread concern to their trustworthiness. In this paper, we study the problem of the Trojan network (TrojanNet) detection in the data-scarce regime, where only the weights of a trained DNN are accessed by the detector. We first propose a data-limited TrojanNet detector (TND), when only a few data samples are available for TrojanNet detection. We show that an effective data-limited TND can be established by exploring connections between Trojan attack and prediction-evasion adversarial attacks including per-sample attack as well as all-sample universal attack. In addition, we propose a data-free TND, which can detect a TrojanNet without accessing any data samples. We show that such a TND can be built by leveraging the internal response of hidden neurons, which exhibits the Trojan behavior even at random noise inputs. The effectiveness of our proposals is evaluated by extensive experiments under different model architectures and datasets including CIFAR-10, GTSRB, and ImageNet.

Effective Inhibition of SARS-CoV-2 Entry by Heparin and Enoxaparin Derivatives

Abstract: Severe acute respiratory syndrome-related coronavirus 2 (SARS-CoV-2) has caused a pandemic of historic proportions and continues to spread globally, with enormous consequences to human health. Currently there is no vaccine, effective therapeutic or prophylactic. Like other betacoronaviruses, attachment and entry of SARS-CoV-2 is mediated by the spike glycoprotein (SGP). In addition to its well-documented interaction with its receptor, human angiotensin converting enzyme 2 (hACE2), SGP has been found to bind to glycosaminoglycans like heparan sulfate, which is found on the surface of virtually all mammalian cells. Here, we pseudotyped SARS-CoV-2 SGP on a third generation lentiviral (pLV) vector and tested the impact of various sulfated polysaccharides on transduction efficiency in mammalian cells. The pLV vector pseudotyped SGP efficiently and produced high titers on HEK293T cells. Various sulfated polysaccharides potently neutralized pLV-S pseudotyped virus with clear structure-based differences in anti-viral activity and affinity to SGP. Concentration-response curves showed that pLV-S particles were efficiently neutralized by a range of concentrations of unfractionated heparin (UFH), enoxaparin, 6-O-desulfated UFH and 6-O-desulfated enoxaparin with an IC50 of 5.99 μg/L, 1.08 mg/L, 1.77 μg/L, and 5.86 mg/L respectively. The low serum bioavailability of intranasally administered UFH, along with data suggesting that the nasal epithelium is a portal for initial infection and transmission, suggest that intranasal administration of UFH may be an effective and safe prophylactic treatment.

Multifunctional Bio‐Nanocomposite Coatings for Perishable Fruits

Abstract: Hunger and chronic undernourishment impact over 800 million people, which translates to ≈10.7% of the world's population. While countries are increasingly making efforts to reduce poverty and hunger by pursuing sustainable energy and agricultural practices, a third of the food produced around the globe still is wasted and never consumed. Reducing food shortages is vital in this effort and is often addressed by the development of genetically modified produce or chemical additives and inedible coatings, which create additional health and environmental concerns. Herein, a multifunctional bio-nanocomposite comprised largely of egg-derived polymers and cellulose nanomaterials as a conformal coating onto fresh produce that slows down food decay by retarding ripening, dehydration, and microbial invasion is reported. The coating is edible, washable, and made from readily available inexpensive or waste materials, which makes it a promising economic alternative to commercially available fruit coatings and a solution to combat food wastage that is rampant in the world.