Showing papers by "University of Colorado Colorado Springs published in 2021"
TL;DR: The field of natural language processing has been propelled forward by an explosion in the use of deep learning models over the last several years as mentioned in this paper, which includes several core linguistic processing issues in addition to many applications of computational linguistics.
Abstract: Over the last several years, the field of natural language processing has been propelled forward by an explosion in the use of deep learning models. This article provides a brief introduction to the field and a quick overview of deep learning architectures and methods. It then sifts through the plethora of recent studies and summarizes a large assortment of relevant contributions. Analyzed research areas include several core linguistic processing issues in addition to many applications of computational linguistics. A discussion of the current state of the art is then provided along with recommendations for future research in the field.
783 citations
TL;DR: This work proposes the first systematic framework for using deep learning to detect vulnerabilities, dubbed Syntax- based, Semantics-based, and Vector Representations (SySeVR), which focuses on obtaining program representations that can accommodate syntax and semantic information pertinent to vulnerabilities.
Abstract: The detection of software vulnerabilities (or vulnerabilities for short) is an important problem that has yet to be tackled, as manifested by the many vulnerabilities reported on a daily basis. This calls for machine learning methods for vulnerability detection. Deep learning is attractive for this purpose because it alleviates the requirement to manually define features. Despite the tremendous success of deep learning in other application domains, its applicability to vulnerability detection is not systematically understood. In order to fill this void, we propose the first systematic framework for using deep learning to detect vulnerabilities in C/C++ programs with source code. The framework, dubbed Syntax-based, Semantics-based, and Vector Representations (SySeVR), focuses on obtaining program representations that can accommodate syntax and semantic information pertinent to vulnerabilities. Our experiments with 4 software products demonstrate the usefulness of the framework: we detect 15 vulnerabilities that are not reported in the National Vulnerability Database. Among these 15 vulnerabilities, 7 are unknown and have been reported to the vendors, and the other 8 have been “silently” patched by the vendors when releasing newer versions of the pertinent software products.
218 citations
Bose Corporation1, Cardiff University2, Durham University3, Adam Mickiewicz University in Poznań4, Max Planck Society5, Katholieke Universiteit Leuven6, Delft University of Technology7, Georgia Institute of Technology8, Kaiserslautern University of Technology9, Beihang University10, École Polytechnique Fédérale de Lausanne11, Saratov State University12, Paris-Sorbonne University13, The Catholic University of America14, University of Notre Dame15, University of Münster16, Emory University17, Polytechnic University of Milan18, Helmholtz-Zentrum Dresden-Rossendorf19, Dresden University of Technology20, University of Exeter21, Donetsk National University22, SRM University23, National University of Singapore24, University of Greifswald25, Kyoto University26, Tohoku University27, University of Santiago, Chile28, Federico Santa María Technical University29, University of Perugia30, Université Paris-Saclay31, University of Manitoba32, University of Colorado Colorado Springs33, University of Tokyo34, University of Groningen35, Technische Universität München36, Technical University of Dortmund37, University of Vienna38, Aalto University39, University of California, Riverside40, Intel41, University of Duisburg-Essen42, University of Oldenburg43
TL;DR: The Roadmap on Magnonics as mentioned in this paper is a collection of 22 sections written by leading experts in this field who review and discuss the current status but also present their vision of future perspectives.
Abstract: Magnonics is a rather young physics research field in nanomagnetism and nanoscience that addresses the use of spin waves (magnons) to transmit, store, and process information. After several papers and review articles published in the last decade, with a steadily increase in the number of citations, we are presenting the first Roadmap on Magnonics. This a collection of 22 sections written by leading experts in this field who review and discuss the current status but also present their vision of future perspectives. Today, the principal challenges in applied magnonics are the excitation of sub-100 nm wavelength magnons, their manipulation on the nanoscale and the creation of sub-micrometre devices using low-Gilbert damping magnetic materials and the interconnections to standard electronics. In this respect, magnonics offers lower energy consumption, easier integrability and compatibility with CMOS structure, reprogrammability, shorter wavelength, smaller device features, anisotropic properties, negative group velocity, non-reciprocity and efficient tunability by various external stimuli to name a few. Hence, despite being a young research field, magnonics has come a long way since its early inception. This Roadmap represents a milestone for future emerging research directions in magnonics and hopefully it will be followed by a series of articles on the same topic.
188 citations
TL;DR: In this article, the authors discuss the implications of awareness of death for understanding the widely varying ways in which awareness plays a role in different aspects of life, and discuss the theory's implications for understanding how people cope with death.
Abstract: Terror management theory is focused on the role that awareness of death plays in diverse aspects of life. Here, we discuss the theory’s implications for understanding the widely varying ways in whi...
173 citations
TL;DR: This double-blind, randomised, parallel-group, phase 3A study investigated the safety and efficacy of single-inhaler fluticasone furoate plus umeclidinium plus vilanterol compared with FF/VI in patients with moderate or severe asthma.
114 citations
TL;DR: In this article, the authors evaluate the scalability of primary, secondary and tertiary interventions according to mental health target, population, modality, intensity and provider type to provide a unified strategy for meeting population mental health needs.
82 citations
TL;DR: The findings reinforce the utility of developing intervention strategies that minimize harm to children by targeting resources to specific challenges facing families enduring the COVID-19 experience.
63 citations
American Society of Clinical Oncology1, University of Texas MD Anderson Cancer Center2, Harvard University3, University of New Mexico Hospital4, Cancer Care Ontario5, University of Colorado Colorado Springs6, Brigham and Women's Hospital7, Fred Hutchinson Cancer Research Center8, Johns Hopkins University9, Stanford University10
TL;DR: Patients with Her2-positive breast cancer with pathologic invasive residual disease at surgery after standard preoperative chemotherapy and HER2-targeted therapy should be offered 14 cycles of adjuvant T-DM1, unless there is disease recurrence or unmanageable toxicity.
Abstract: PURPOSEThe aim of this work is to update key recommendations of the ASCO guideline adaptation of the Cancer Care Ontario guideline on the selection of optimal adjuvant chemotherapy regimens for ear...
54 citations
TL;DR: In this paper, the authors provide marketing managers with insights into how businesses pursuing various strategies approach these digital marketing issues, with the ultimate goal being to assist managers in the efficient and effective implementation of their firm's adopted strategy.
52 citations
TL;DR: The mental health risks that the authors observed early in the COVID-19 pandemic are elevated above previous viral outbreaks (SARS) and comparable to rates shown in disasters (9/11 attacks; Hurricane Katrina); there was an inverse relationship between number of positive CO VID-19 cases and anxiety, such that as positive cases went up, anxiety decreased.
46 citations
Martin Schweinsberg1, Michael Feldman2, Nicola Staub2, Olmo van den Akker3 +175 more•Institutions (121)
TL;DR: DataExplained as discussed by the authors is a crowdsourced initiative that allows independent analysts to test two hypotheses regarding the effects of scientists' gender and professional status on verbosity during group meetings using the same dataset.
University of Maryland, Baltimore1, Silver Spring Networks2, University of British Columbia3, University of Texas Health Science Center at Houston4, University of Manitoba5, University of Chicago6, Stanford University7, Queen's University8, East Carolina University9, University of Western Ontario10, University of Texas Medical Branch11, Baystate Medical Center12, Vancouver Island Health Authority13, MedStar Union Memorial Hospital14, Université de Montréal15, Sunnybrook Health Sciences Centre16, Yale University17, Franciscan Health18, Cornell University19, Vanderbilt University Medical Center20, Brigham and Women's Hospital21, University of Calgary22, Beth Israel Deaconess Medical Center23, California State University, Long Beach24, University of California, Davis25, Georgia Regents University26, Dalhousie University27, University of Alberta28, Northwestern University29, University of Colorado Colorado Springs30, Washington University in St. Louis31, Toronto General Hospital32, University of Texas Southwestern Medical Center33, Montreal Heart Institute34, The Texas Heart Institute35, Baylor College of Medicine36, Houston Methodist Hospital37, University of Kentucky38, Brown University39, Mayo Clinic40, Lankenau Medical Center41, Michigan State University42, Johns Hopkins University School of Medicine43, University of Toronto44, Medical University of South Carolina45
TL;DR: The first wave of the COVID-19 pandemic resulted in an increase in hospital resource utilization and the need to defer non-urgent cardiac surgery procedures, resulting in a decline to 45% of baseline with significant regional variation.
Book•
21 Oct 2021TL;DR: A Genealogy of Dissent as mentioned in this paper traces the history of these progressive Baptists and their descendants throughout the twentieth century and shows how they created an active culture of protest within a highly traditional society.
Abstract: " A Choice Outstanding Academic Title Between the Civil War and the turn of the last century, Southern Baptists gained prominence in the religious life of the South. As their power increased, they became defenders of the racial, political, social, and economic status quo. By the beginning of this century, however, a feisty tradition of dissent began to appear in Southern Baptist life as criticism of the center increased from both the left and the right. The popular belief in a doctrine of ""once saved, always saved"" led progressive Baptists to claim that moderates, once saved, did not address the serious social and political problems that faced many in the South. These Baptist dissenters claimed that they could not be ""at ease in Zion."" Led by the radical Walter Nathan Johnson in the 1920s and 1930s, progressive Baptists produced civil rights advocates, labor organizers, women's rights advocates, and proponents of disarmament and abolition of capital punishment. They challenged some of the most fundamental aspects of southern society and of Baptist ecclesiastical structure and practice. For their efforts and beliefs, many of these men and women suffered as they lost jobs, experienced physical danger and injury, and endured character assassination. In A Genealogy of Dissent, David Stricklin traces the history of these progressive Baptists and their descendants throughout the twentieth century and shows how they created an active culture of protest within a highly traditional society.
TL;DR: The framework may be considered as a basis for extending the effect of uncertainty on consequential performance, as opposed to assuming uncertainty will impede effectiveness or testing uncertainty and managerial interventions as separate contingencies.
01 Jun 2021
TL;DR: The Zillow Indoor Dataset (ZInD) as discussed by the authors is a large indoor dataset with 71,474 panoramas from 1,524 real unfurnished homes and provides annotations of 3D room layouts, 2D and 3D floor plans, panorama location in the floor plan, and locations of windows and doors.
Abstract: We present Zillow Indoor Dataset (ZInD): A large indoor dataset with 71,474 panoramas from 1,524 real unfurnished homes. ZInD provides annotations of 3D room layouts, 2D and 3D floor plans, panorama location in the floor plan, and locations of windows and doors. The ground truth construction took over 1,500 hours of annotation work. To the best of our knowledge, ZInD is the largest real dataset with layout annotations. A unique property is the room layout data, which follows a real world distribution (cuboid, more general Manhattan, and non-Manhattan layouts) as opposed to the mostly cuboid or Manhattan layouts in current publicly available datasets. Also, the scale and annotations provided are valuable for effective research related to room layout and floor plan analysis. To demonstrate ZInD’s benefits, we benchmark on room layout estimation from single panoramas and multi-view registration.
TL;DR: This randomized clinical equivalence trial found that SB11 demonstrated equivalence in efficacy for both primary end points: adjusted treatment differences between groups were within predefined equivalence margins.
Abstract: Importance Neovascular age-related macular degeneration is the leading cause of blindness in individuals 50 years or older. The availability of a ranibizumab biosimilar product (SB11) may facilitate access to an effective alternative to this treatment. Objective To demonstrate equivalence of efficacy, similar safety, and similar immunogenicity of SB11 compared with the reference ranibizumab. Design, Setting, and Participants This randomized, double-masked, parallel-group phase 3 equivalence study was conducted in 75 centers in 9 countries from March 14, 2018, to December 9, 2019, among 705 participants 50 years or older with neovascular age-related macular degeneration with active subfoveal choroidal neovascularization lesions. Analysis was performed on an intent-to-treat basis. Interventions Intravitreous injection of SB11 or ranibizumab, 0.5 mg, every 4 weeks through week 48. Main Outcomes and Measures Preplanned interim analysis after all participants completed the week 24 assessment of primary efficacy end points at week 8 for change from baseline in best-corrected visual acuity (BCVA) and week 4 for central subfield thickness (CST), with predefined equivalence margins for adjusted treatment differences of −3 letters to 3 letters for BCVA and −36 μm to 36 μm for CST. Results Baseline and disease characteristics among 705 randomized participants (403 women [57.2%]; mean [SD] age, 74.1 [8.5] years) were comparable between treatment groups (SB11, 351; ranibizumab, 354). Least-squares mean (SE) changes in BCVA from baseline at week 8 were 6.2 (0.5) letters in the SB11 group vs 7.0 (0.5) letters in the ranibizumab group, with an adjusted treatment difference of −0.8 letter (90% CI, −1.8 to 0.2 letters). Least-squares mean (SE) changes in CST from baseline at week 4 were −108 (5) μm in the SB11 group vs −100 (5) μm in the ranibizumab group, with an adjusted treatment difference of −8 μm (95% CI, −19 to 3 μm). Incidences of treatment-emergent adverse events (231 of 350 [66.0%] vs 237 of 354 [66.9%]), including serious treatment-emergent adverse events (44 of 350 [12.6%] vs 44 of 354 [12.4%]) and treatment-emergent adverse events leading to study drug discontinuation (8 of 350 [2.3%] vs 5 of 354 [1.4%]), were similar in the SB11 and ranibizumab groups. Immunogenicity was low, with a cumulative incidence of antidrug antibodies up to week 24 of 3.0% (10 of 330) in the SB11 group and 3.1% (10 of 327) in the ranibizumab group. Conclusions and Relevance These findings of equivalent efficacy and similar safety and immunogenicity profiles compared with ranibizumab support the use of SB11 for patients with neovascular age-related macular degeneration. Trial Registration ClinicalTrials.gov Identifier:NCT03150589
TL;DR: Lipedema is a loose connective tissue disease predominantly in women identified by increased nodular and fibrotic adipose tissue on the buttocks, hips and limbs that develops at times of...
Abstract: BackgroundLipedema is a loose connective tissue disease predominantly in women identified by increased nodular and fibrotic adipose tissue on the buttocks, hips and limbs that develops at times of ...
University of California, Los Angeles1, New York University2, University of Melbourne3, University of Oregon4, University of South Florida5, Cornell University6, Pennsylvania State University7, University of California, Berkeley8, University of Kansas9, Columbia University10, University of Kent11, University of California, Davis12, Duke University13, University of New Mexico14, University of North Carolina at Chapel Hill15, William Paterson University16, Brock University17, Claremont Graduate University18, University of Queensland19, University of Western Ontario20, Boston College21, University of Edinburgh22, University of Calgary23, University of Minnesota24, Tulane University25, University of Colorado Colorado Springs26, Bellarmine University27, University of British Columbia28, University of California, Irvine29, University of Virginia30, University of Pennsylvania31
TL;DR: The COVID-19 pandemic has extensively changed the state of psychological science from what research questions psychologists can ask to which methodologies psychologists can use to investigate them as mentioned in this paper, and the authors offer a perspective on how to optimize new research in the pandemic's wake.
Abstract: The COVID-19 pandemic has extensively changed the state of psychological science from what research questions psychologists can ask to which methodologies psychologists can use to investigate them. In this article, we offer a perspective on how to optimize new research in the pandemic's wake. Because this pandemic is inherently a social phenomenon-an event that hinges on human-to-human contact-we focus on socially relevant subfields of psychology. We highlight specific psychological phenomena that have likely shifted as a result of the pandemic and discuss theoretical, methodological, and practical considerations of conducting research on these phenomena. After this discussion, we evaluate metascientific issues that have been amplified by the pandemic. We aim to demonstrate how theoretically grounded views on the COVID-19 pandemic can help make psychological science stronger-not weaker-in its wake.
TL;DR: Results suggest the ACS WCT has utility for detecting invalid performance in a clinical sample with likely cognitive impairment, though the embedded ACS measures may have limited incremental utility, particularly in individuals with cognitive impairment.
Abstract: Use of multiple performance validity tests (PVTs) may best identify invalid performance, though few studies have examined the utility and accuracy of combining PVTs. This study examined the followi...
TL;DR: In this article, the authors highlight the importance of cultural diversity as a source of resilience in the past and apply the knowledge gained to the present, contributing a much needed, long-term perspective to climate research.
Abstract: Anthropogenic climate change is currently driving environmental transformation on a scale and at a pace that exceeds historical records. This represents an undeniably serious challenge to existing social, political, and economic systems. Humans have successfully faced similar challenges in the past, however. The archaeological record and Earth archives offer rare opportunities to observe the complex interaction between environmental and human systems under different climate regimes and at different spatial and temporal scales. The archaeology of climate change offers opportunities to identify the factors that promoted human resilience in the past and apply the knowledge gained to the present, contributing a much-needed, long-term perspective to climate research. One of the strengths of the archaeological record is the cultural diversity it encompasses, which offers alternatives to the solutions proposed from within the Western agro-industrial complex, which might not be viable cross-culturally. While contemporary climate discourse focuses on the importance of biodiversity, we highlight the importance of cultural diversity as a source of resilience.
TL;DR: There was no overall increase in incidence of glomerular disease when compared with the 2 years prior to the COVID-19 pandemic diagnosed on kidney biopsies in this practice, although it should be monitored as a potential adverse event.
Abstract: Background: Immune responses to vaccination are a known trigger for a new onset of glomerular disease or disease flare in susceptible individuals. Mass immunization against SARS-CoV-2 in the COVID-19 pandemic provides a unique opportunity to study vaccination-associated autoimmune kidney diseases. In the recent literature, there are several case reports demonstrating a temporal association of SARS-CoV-2 immunization and kidney diseases. Methods: Here, we present a series of 29 cases of biopsy-proven glomerular disease in patients recently vaccinated against SARS-CoV-2 and identified patients who developed a new onset of IgA nephropathy, minimal change disease, membranous nephropathy, ANCA-associated glomerulonephritis, collapsing glomerulopathy, and diffuse lupus nephritis diagnosed on kidney biopsies post-immunization, as well as recurrent ANCA-associated glomerulonephritis. This included 28 cases of de novo glomerulonephritis within native kidney biopsies and one disease flare in an allograft. Results: The patients with collapsing glomerulopathy were of African American descent and had two APOL1 genomic risk alleles. A brief literature review of case reports and small series is also provided to include all reported cases to date (n=52). The incidence of induction of glomerular disease in response to SARS-CoV-2 immunization is unknown, however, there was no overall increase in incidence of glomerular disease when compared to the two years prior to the COVID-19 pandemic diagnosed on kidney biopsies in our practice. Conclusions: This suggests that glomerulonephritis in response to vaccination is rare, although should be monitored as a potential adverse event.
TL;DR: In this article, the authors examined how employee perceptions of corporate social responsibility (CSR) influenced their affective organizational commitment (AOC) via procedural justice and organization-based self-esteem (OBSE).
TL;DR: Online education is a pedagogical innovation that often represents a significant challenge for faculty in most, if not all, disciplines in higher education (Mansbach & Austin, 2018) as mentioned in this paper.
Abstract: Online education is a pedagogical innovation that often represents a significant challenge for faculty in most, if not all, disciplines in higher education (Mansbach & Austin, 2018). Over time, sev...
TL;DR: VulDeeLocator as discussed by the authors is a deep learning-based location-based vulnerability detector that can simultaneously achieve a high detection capability and a high locating precision, dubbed Vulnerability Deep Learning-based Locator.
Abstract: Automatically detecting software vulnerabilities is an important problem that has attracted much attention from the academic research community. However, existing vulnerability detectors still cannot achieve the vulnerability detection capability and the locating precision that would warrant their adoption for real-world use. In this paper, we present a vulnerability detector that can simultaneously achieve a high detection capability and a high locating precision, dubbed Vulnerability Deep learning-based Locator (VulDeeLocator).In the course of designing VulDeeLocator, we encounter difficulties including how to accommodate semantic relations between the definitions of types as well as macros and their uses across files, how to accommodate accurate control flows and variable define-use relations, and how to achieve high locating precision. We solve these difficulties by using two innovative ideas: (i) leveraging intermediate code to accommodate extra semantic information, and (ii) using the notion of granularity refinement to pin down locations of vulnerabilities. When applied to 200 files randomly selected from three real-world software products, VulDeeLocator detects 18 confirmed vulnerabilities (i.e., true-positives). Among them, 16 vulnerabilities correspond to known vulnerabilities; the other two are not reported in the National Vulnerability Database (NVD) but have been silently patched by the vendor of Libav when releasing newer versions.
University of Washington1, University of Alabama at Birmingham2, Fred Hutchinson Cancer Research Center3, University of North Carolina at Chapel Hill4, University of Texas Health Science Center at Houston5, Tulane University6, University of Hamburg7, Boston University8, Central Michigan University9, University of Western Ontario10, University of Colorado Colorado Springs11, Brigham and Women's Hospital12, National Institutes of Health13, University of Mississippi Medical Center14, Broad Institute15, UCLA Medical Center16, Yale University17, Ohio State University18, Johns Hopkins University19, Brown University20, University of Maryland, College Park21, University of Pittsburgh22, Columbia University23, University of Minnesota24, VCU Medical Center25
TL;DR: In this paper, the authors describe a centralized system for phenotype harmonization that includes input from phenotype domain and study experts, quality control, documentation, reproducible results, and data-sharing mechanisms.
Abstract: Genotype-phenotype association studies often combine phenotype data from multiple studies to increase statistical power. Harmonization of the data usually requires substantial effort due to heterogeneity in phenotype definitions, study design, data collection procedures, and data-set organization. Here we describe a centralized system for phenotype harmonization that includes input from phenotype domain and study experts, quality control, documentation, reproducible results, and data-sharing mechanisms. This system was developed for the National Heart, Lung, and Blood Institute's Trans-Omics for Precision Medicine (TOPMed) program, which is generating genomic and other -omics data for more than 80 studies with extensive phenotype data. To date, 63 phenotypes have been harmonized across thousands of participants (recruited in 1948-2012) from up to 17 studies per phenotype. Here we discuss challenges in this undertaking and how they were addressed. The harmonized phenotype data and associated documentation have been submitted to National Institutes of Health data repositories for controlled access by the scientific community. We also provide materials to facilitate future harmonization efforts by the community, which include 1) the software code used to generate the 63 harmonized phenotypes, enabling others to reproduce, modify, or extend these harmonizations to additional studies, and 2) the results of labeling thousands of phenotype variables with controlled vocabulary terms.
TL;DR: For example, Bloodgood et al. as discussed by the authors investigated the influence of sex and age on candidate performance before academy training and found that women and older candidates may have a lower physical capacity than men and younger candidates, respectively, in certain physical assessments.
Abstract: Bloodgood, AM, Dawes, JJ, Orr, RM, Stierli, M, Cesario, KA, Moreno, MR, Dulla, JM, and Lockie, RG. Effects of sex and age on physical testing performance for law enforcement agency candidates: implications for academy training. J Strength Cond Res 35(9): 2629-2635, 2021-Law enforcement candidates are often required to complete fitness tests to assess their preparedness for training. This study investigated the influence of sex and age on candidate performance before academy training. Retrospective analysis of 516 candidates was performed. Data were stratified into men (n = 432) and women (n = 84), and the pooled (men and women combined) data were stratified into age groups (20-24; 24-29; 30-34; 35-39; and 40+ years). The tests included the following: maximal push-ups and sit-ups in 60 seconds; a 75-yard pursuit run (75PR); arm ergometer revolutions completed in 60 seconds; and a 2.4-km run. To compare the sex and age groups, a 2 × 5 analysis of variance (with Bonferroni post hoc for multiple between-age group comparisons) was used. Men performed significantly (p < 0.001) more push-ups, sit-ups, and arm ergometer revolutions and were faster in the 75PR and 2.4-km run. The 20-24, 25-29, and 35-39 year groups were faster in the 75PR compared with the 40+ year group; the 20-24 and 25-29 year groups were faster than the 35-39 year group; and the 20-24 year group was faster than the 30-34 year group (p ≤ 0.023). The 20-24 year group was faster than the 40+ year group in the 2.4-km run (p = 0.005). Women and older candidates may have a lower physical capacity than men and younger candidates, respectively, in certain physical assessments. Women and older candidates, in particular, may benefit from dedicated training before academy to better prepare for the demands of this period.
TL;DR: AutoMeKin2021 as mentioned in this paper is an updated version of tsscds2018, a program for the automated discovery of reaction mechanisms (J. Comput. Chem. 2018, 39, 1922).
Abstract: AutoMeKin2021 is an updated version of tsscds2018, a program for the automated discovery of reaction mechanisms (J. Comput. Chem. 2018, 39, 1922). This release features a number of new capabilities: rare-event molecular dynamics simulations to enhance reaction discovery, extension of the original search algorithm to study van der Waals complexes, use of chemical knowledge, a new search algorithm based on bond-order time series analysis, statistics of the chemical reaction networks, a web application to submit jobs, and other features. The source code, manual, installation instructions and the website link are available at: https://rxnkin.usc.es/index.php/AutoMeKin.
01 Apr 2021
TL;DR: In this article, the effect of the COVID-19 pandemic on the distribution and determinants of traumatic injuries has been investigated in six community level I trauma centers in the USA.
Abstract: Background The COVID-19 pandemic resulted in nationwide social distancing and shelter-in-place orders meant to curb transmission of the SARS-CoV-2 virus. The effect of the pandemic on injury patterns has not been well described in the USA. The study objective is to determine the effect of the COVID-19 pandemic on the distribution and determinants of traumatic injuries. Methods This retrospective multi-institutional cohort study included all hospital admissions for acute traumatic injury at six community level I trauma centers. Descriptive statistics were used to compare injury causes, diagnoses and procedures over two similar time periods: prepandemic (March 11–June 30, 2019) and pandemic (March 11–June 30, 2020). Results There were 7308 trauma patients included: 3862 (53%) prepandemic and 3446 (47%) during the pandemic. Cause of injury significantly differed by period (p Discussion Causes of traumatic injury, diagnoses, and procedures were significantly changed by the pandemic. Trauma centers must adjust to meet the changing demands associated with altered injury patterns, as they were associated with increased use of hospital resources. Level of evidence III (epidemiological).
TL;DR: iPart as discussed by the authors partitions communication and computation in various partition sizes to overlap gradient communication with backward computation and parameter communication with forward computation to reduce the impact of communication constraint on the scalability.
Abstract: Scalability of distributed deep learning (DL) training with parameter server (PS) architecture is often communication constrained in large clusters. There are recent efforts that use a layer by layer strategy to overlap gradient communication with backward computation so as to reduce the impact of communication constraint on the scalability. However, the approaches could bring significant overhead in gradient communication. Meanwhile, they cannot be effectively applied to the overlap between parameter communication and forward computation. In this article, we propose and develop iPart, a novel approach that partitions communication and computation in various partition sizes to overlap gradient communication with backward computation and parameter communication with forward computation. iPart formulates the partitioning decision as an optimization problem and solves it based on a greedy algorithm to derive communication and computation partitions. We implement iPart in the open-source DL framework BigDL and perform evaluations with various DL workloads. Experimental results show that iPart improves the scalability of a cluster of 72 nodes by up to 94 percent over the default PS and 52 percent over the layer by layer strategy.