Showing papers by "New York University published in 2014"
••
06 Sep 2014TL;DR: A novel visualization technique is introduced that gives insight into the function of intermediate feature layers and the operation of the classifier in large Convolutional Network models, used in a diagnostic role to find model architectures that outperform Krizhevsky et al on the ImageNet classification benchmark.
Abstract: Large Convolutional Network models have recently demonstrated impressive classification performance on the ImageNet benchmark Krizhevsky et al. [18]. However there is no clear understanding of why they perform so well, or how they might be improved. In this paper we explore both issues. We introduce a novel visualization technique that gives insight into the function of intermediate feature layers and the operation of the classifier. Used in a diagnostic role, these visualizations allow us to find model architectures that outperform Krizhevsky et al on the ImageNet classification benchmark. We also perform an ablation study to discover the performance contribution from different model layers. We show our ImageNet model generalizes well to other datasets: when the softmax classifier is retrained, it convincingly beats the current state-of-the-art results on Caltech-101 and Caltech-256 datasets.
12,783 citations
••
25 Aug 2014TL;DR: The CNN models discussed herein improve upon the state of the art on 4 out of 7 tasks, which include sentiment analysis and question classification, and are proposed to allow for the use of both task-specific and static vectors.
Abstract: We report on a series of experiments with convolutional neural networks (CNN) trained on top of pre-trained word vectors for sentence-level classification tasks. We show that a simple CNN with little hyperparameter tuning and static vectors achieves excellent results on multiple benchmarks. Learning task-specific vectors through fine-tuning offers further gains in performance. We additionally propose a simple modification to the architecture to allow for the use of both task-specific and static vectors. The CNN models discussed herein improve upon the state of the art on 4 out of 7 tasks, which include sentiment analysis and question classification.
9,776 citations
•
01 Jan 2014TL;DR: It is found that there is no distinction between individual highlevel units and random linear combinations of high level units, according to various methods of unit analysis, and it is suggested that it is the space, rather than the individual units, that contains of the semantic information in the high layers of neural networks.
Abstract: Deep neural networks are highly expressive models that have recently achieved state of the art performance on speech and visual recognition tasks. While their expressiveness is the reason they succeed, it also causes them to learn uninterpretable solutions that could have counter-intuitive properties. In this paper we report two such properties.
First, we find that there is no distinction between individual high level units and random linear combinations of high level units, according to various methods of unit analysis. It suggests that it is the space, rather than the individual units, that contains of the semantic information in the high layers of neural networks.
Second, we find that deep neural networks learn input-output mappings that are fairly discontinuous to a significant extend. We can cause the network to misclassify an image by applying a certain imperceptible perturbation, which is found by maximizing the network's prediction error. In addition, the specific nature of these perturbations is not a random artifact of learning: the same perturbation can cause a different network, that was trained on a different subset of the dataset, to misclassify the same input.
9,561 citations
••
University of Washington1, Sapienza University of Rome2, Mekelle University3, University of Texas at San Antonio4, King Saud bin Abdulaziz University for Health Sciences5, Debre markos University6, Emory University7, University of Oxford8, University of Cartagena9, United Nations Population Fund10, University of Birmingham11, Stanford University12, Aga Khan University13, University of Melbourne14, National Taiwan University15, University of Cambridge16, University of California, San Diego17, Public Health Foundation of India18, Public Health England19, University of Peradeniya20, Harvard University21, National Institutes of Health22, Tehran University of Medical Sciences23, Auckland University of Technology24, University of Sheffield25, University of Western Australia26, Karolinska Institutet27, Birzeit University28, Brandeis University29, American Cancer Society30, Ochsner Medical Center31, Yonsei University32, University of Bristol33, Heidelberg University34, Vanderbilt University35, South African Medical Research Council36, Jordan University of Science and Technology37, New Generation University College38, Northeastern University39, Simmons College40, Norwegian Institute of Public Health41, Boston University42, Chinese Center for Disease Control and Prevention43, University of Bari44, University of São Paulo45, University of Otago46, University of Crete47, International Centre for Diarrhoeal Disease Research, Bangladesh48, Fred Hutchinson Cancer Research Center49, Teikyo University50, Bhabha Atomic Research Centre51, University of Tokyo52, Finnish Institute of Occupational Health53, Heriot-Watt University54, University of Alabama at Birmingham55, Griffith University56, National Center for Disease Control and Public Health57, University of California, Irvine58, Johns Hopkins University59, New York University60, University of Queensland61, Universidade Federal de Minas Gerais62, National Research University – Higher School of Economics63, University of Bergen64, Columbia University65, Shandong University66, University of North Carolina at Chapel Hill67, Fujita Health University68, Korea University69, Chongqing Medical University70, Zhejiang University71
TL;DR: The global, regional, and national prevalence of overweight and obesity in children and adults during 1980-2013 is estimated using a spatiotemporal Gaussian process regression model to estimate prevalence with 95% uncertainty intervals (UIs).
9,180 citations
•
TL;DR: In this article, CNNs are trained on top of pre-trained word vectors for sentence-level classification tasks and a simple CNN with little hyperparameter tuning and static vectors achieves excellent results on multiple benchmarks.
Abstract: We report on a series of experiments with convolutional neural networks (CNN) trained on top of pre-trained word vectors for sentence-level classification tasks. We show that a simple CNN with little hyperparameter tuning and static vectors achieves excellent results on multiple benchmarks. Learning task-specific vectors through fine-tuning offers further gains in performance. We additionally propose a simple modification to the architecture to allow for the use of both task-specific and static vectors. The CNN models discussed herein improve upon the state of the art on 4 out of 7 tasks, which include sentiment analysis and question classification.
7,826 citations
••
University of Iowa1, University of Alabama at Birmingham2, University of Tennessee Health Science Center3, Johns Hopkins University4, Kaiser Permanente5, Medical University of South Carolina6, University of Missouri7, University of Colorado Denver8, New York University9, University of North Carolina at Chapel Hill10, Duke University11, Mayo Clinic12, University of Pennsylvania13, Case Western Reserve University14, National Institutes of Health15
TL;DR: Although this guideline provides evidence-based recommendations for the management of high BP and should meet the clinical needs of most patients, these recommendations are not a substitute for clinical judgment, and decisions about care must carefully consider and incorporate the clinical characteristics and circumstances of each individual patient.
Abstract: Hypertension is the most common condition seen in primary care and leads to myocardial infarction, stroke, renal failure, and death if not detected early and treated appropriately. Patients want to be assured that blood pressure (BP) treatment will reduce their disease burden, while clinicians want guidance on hypertension management using the best scientific evidence. This report takes a rigorous, evidence-based approach to recommend treatment thresholds, goals, and medications in the management of hypertension in adults. Evidence was drawn from randomized controlled trials, which represent the gold standard for determining efficacy and effectiveness. Evidence quality and recommendations were graded based on their effect on important outcomes. There is strong evidence to support treating hypertensive persons aged 60 years or older to a BP goal of less than 150/90 mm Hg and hypertensive persons 30 through 59 years of age to a diastolic goal of less than 90 mm Hg; however, there is insufficient evidence in hypertensive persons younger than 60 years for a systolic goal, or in those younger than 30 years for a diastolic goal, so the panel recommends a BP of less than 140/90 mm Hg for those groups based on expert opinion. The same thresholds and goals are recommended for hypertensive adults with diabetes or nondiabetic chronic kidney disease (CKD) as for the general hypertensive population younger than 60 years. There is moderate evidence to support initiating drug treatment with an angiotensin-converting enzyme inhibitor, angiotensin receptor blocker, calcium channel blocker, or thiazide-type diuretic in the nonblack hypertensive population, including those with diabetes. In the black hypertensive population, including those with diabetes, a calcium channel blocker or thiazide-type diuretic is recommended as initial therapy. There is moderate evidence to support initial or add-on antihypertensive therapy with an angiotensin-converting enzyme inhibitor or angiotensin receptor blocker in persons with CKD to improve kidney outcomes. Although this guideline provides evidence-based recommendations for the management of high BP and should meet the clinical needs of most patients, these recommendations are not a substitute for clinical judgment, and decisions about care must carefully consider and incorporate the clinical characteristics and circumstances of each individual patient.
7,119 citations
••
TL;DR: The Reactome Knowledgebase provides molecular details of signal transduction, transport, DNA replication, metabolism and other cellular processes as an ordered network of molecular transformations—an extended version of a classic metabolic map, in a single consistent data model.
Abstract: The Reactome Knowledgebase (www.reactome.org) provides molecular details of signal transduction, transport, DNA replication, metabolism and other cellular processes as an ordered network of molecular transformations-an extended version of a classic metabolic map, in a single consistent data model. Reactome functions both as an archive of biological processes and as a tool for discovering unexpected functional relationships in data such as gene expression pattern surveys or somatic mutation catalogues from tumour cells. Over the last two years we redeveloped major components of the Reactome web interface to improve usability, responsiveness and data visualization. A new pathway diagram viewer provides a faster, clearer interface and smooth zooming from the entire reaction network to the details of individual reactions. Tool performance for analysis of user datasets has been substantially improved, now generating detailed results for genome-wide expression datasets within seconds. The analysis module can now be accessed through a RESTFul interface, facilitating its inclusion in third party applications. A new overview module allows the visualization of analysis results on a genome-wide Reactome pathway hierarchy using a single screen page. The search interface now provides auto-completion as well as a faceted search to narrow result lists efficiently.
5,065 citations
••
St. Jude Children's Research Hospital1, University of Edinburgh2, Singapore Immunology Network3, New York University4, University College London5, Heidelberg University6, University of Oxford7, Royal Melbourne Hospital8, Hospital for Special Surgery9, University of Milan10, Aix-Marseille University11, University of Maryland, College Park12, European Institute of Oncology13, Massachusetts Institute of Technology14, University of Bonn15, University of Maryland, Baltimore16, University of Eastern Piedmont17, University of Louisville18, Vrije Universiteit Brussel19, National Institutes of Health20
TL;DR: A set of standards encompassing three principles-the source of macrophages, definition of the activators, and a consensus collection of markers to describe macrophage activation are described with the goal of unifying experimental standards for diverse experimental scenarios.
4,287 citations
••
Eric A. Collisson1, Joshua D. Campbell2, Angela N. Brooks2, Angela N. Brooks3 +315 more•Institutions (41)
TL;DR: In this paper, the authors report molecular profiling of 230 resected lung adnocarcinomas using messenger RNA, microRNA and DNA sequencing integrated with copy number, methylation and proteomic analyses.
Abstract: Adenocarcinoma of the lung is the leading cause of cancer death worldwide. Here we report molecular profiling of 230 resected lung adenocarcinomas using messenger RNA, microRNA and DNA sequencing integrated with copy number, methylation and proteomic analyses. High rates of somatic mutation were seen (mean 8.9 mutations per megabase). Eighteen genes were statistically significantly mutated, including RIT1 activating mutations and newly described loss-of-function MGA mutations which are mutually exclusive with focal MYC amplification. EGFR mutations were more frequent in female patients, whereas mutations in RBM10 were more common in males. Aberrations in NF1, MET, ERBB2 and RIT1 occurred in 13% of cases and were enriched in samples otherwise lacking an activated oncogene, suggesting a driver role for these events in certain tumours. DNA and mRNA sequence from the same tumour highlighted splicing alterations driven by somatic genomic changes, including exon 14 skipping in MET mRNA in 4% of cases. MAPK and PI(3)K pathway activity, when measured at the protein level, was explained by known mutations in only a fraction of cases, suggesting additional, unexplained mechanisms of pathway activation. These data establish a foundation for classification and further investigations of lung adenocarcinoma molecular pathogenesis.
4,104 citations
••
Stanford University1, HCL Technologies2, Universidad Mayor3, UCL Institute of Child Health4, University of Bonn5, University of California, Los Angeles6, Umeå University7, New York University8, Columbia University9, Yonsei University10, Albert Einstein College of Medicine11, University of Pavia12, University of Melbourne13, Karolinska Institutet14, University of Calgary15
TL;DR: A revised definition of epilepsy brings the term in concordance with common use for individuals who either had an age‐dependent epilepsy syndrome but are now past the applicable age or who have remained seizure‐free for the last 10 years and off antiseizure medicines for at least the last 5 years.
Abstract: Epilepsy was defined conceptually in 2005 as a disorder of the brain characterized by an enduring predisposition to generate epileptic seizures. This definition is usually practically applied as having two unprovoked seizures >24 h apart. The International League Against Epilepsy (ILAE) accepted recommendations of a task force altering the practical definition for special circumstances that do not meet the two unprovoked seizures criteria. The task force proposed that epilepsy be considered to be a disease of the brain defined by any of the following conditions: (1) At least two unprovoked (or reflex) seizures occurring >24 h apart; (2) one unprovoked (or reflex) seizure and a probability of further seizures similar to the general recurrence risk (at least 60%) after two unprovoked seizures, occurring over the next 10 years; (3) diagnosis of an epilepsy syndrome. Epilepsy is considered to be resolved for individuals who either had an age-dependent epilepsy syndrome but are now past the applicable age or who have remained seizure-free for the last 10 years and off antiseizure medicines for at least the last 5 years. "Resolved" is not necessarily identical to the conventional view of "remission or "cure." Different practical definitions may be formed and used for various specific purposes. This revised definition of epilepsy brings the term in concordance with common use. A PowerPoint slide summarizing this article is available for download in the Supporting Information section here.
3,491 citations
•
21 May 2014TL;DR: This paper considers possible generalizations of CNNs to signals defined on more general domains without the action of a translation group, and proposes two constructions, one based upon a hierarchical clustering of the domain, and another based on the spectrum of the graph Laplacian.
Abstract: Convolutional Neural Networks are extremely efficient architectures in image and audio recognition tasks, thanks to their ability to exploit the local translational invariance of signal classes over their domain. In this paper we consider possible generalizations of CNNs to signals defined on more general domains without the action of a translation group. In particular, we propose two constructions, one based upon a hierarchical clustering of the domain, and another based on the spectrum of the graph Laplacian. We show through experiments that for low-dimensional graphs it is possible to learn convolutional layers with a number of parameters independent of the input size, resulting in efficient deep architectures.
•
08 Dec 2014TL;DR: In this article, two deep network stacks are employed to make a coarse global prediction based on the entire image, and another to refine this prediction locally, which achieves state-of-the-art results on both NYU Depth and KITTI.
Abstract: Predicting depth is an essential component in understanding the 3D geometry of a scene. While for stereo images local correspondence suffices for estimation, finding depth relations from a single image is less straightforward, requiring integration of both global and local information from various cues. Moreover, the task is inherently ambiguous, with a large source of uncertainty coming from the overall scale. In this paper, we present a new method that addresses this task by employing two deep network stacks: one that makes a coarse global prediction based on the entire image, and another that refines this prediction locally. We also apply a scale-invariant error to help measure depth relations rather than scale. By leveraging the raw datasets as large sources of training data, our method achieves state-of-the-art results on both NYU Depth and KITTI, and matches detailed depth boundaries without the need for superpixelation.
••
05 Feb 2014TL;DR: Measurements and capacity studies are surveyed to assess mmW technology with a focus on small cell deployments in urban environments and it is shown that mmW systems can offer more than an order of magnitude increase in capacity over current state-of-the-art 4G cellular networks at current cell densities.
Abstract: Millimeter-wave (mmW) frequencies between 30 and 300 GHz are a new frontier for cellular communication that offers the promise of orders of magnitude greater bandwidths combined with further gains via beamforming and spatial multiplexing from multielement antenna arrays. This paper surveys measurements and capacity studies to assess this technology with a focus on small cell deployments in urban environments. The conclusions are extremely encouraging; measurements in New York City at 28 and 73 GHz demonstrate that, even in an urban canyon environment, significant non-line-of-sight (NLOS) outdoor, street-level coverage is possible up to approximately 200 m from a potential low-power microcell or picocell base station. In addition, based on statistical channel models from these measurements, it is shown that mmW systems can offer more than an order of magnitude increase in capacity over current state-of-the-art 4G cellular networks at current cell densities. Cellular systems, however, will need to be significantly redesigned to fully achieve these gains. Specifically, the requirement of highly directional and adaptive transmissions, directional isolation between links, and significant possibilities of outage have strong implications on multiple access, channel structure, synchronization, and receiver design. To address these challenges, the paper discusses how various technologies including adaptive beamforming, multihop relaying, heterogeneous network architectures, and carrier aggregation can be leveraged in the mmW context.
••
Icahn School of Medicine at Mount Sinai1, Cleveland Clinic2, University of Alabama at Birmingham3, University of Copenhagen4, University College London5, University of Texas Health Science Center at Houston6, New York University7, University of Pennsylvania8, VU University Amsterdam9, National Multiple Sclerosis Society10, Johns Hopkins University11, Vita-Salute San Raffaele University12, University of Ottawa13, University of Rochester14, University of Basel15, University of Düsseldorf16, Pierre-and-Marie-Curie University17, Autonomous University of Barcelona18, University of Toronto19, University of British Columbia20, Sapienza University of Rome21, University of Texas Southwestern Medical Center22, University of California, San Francisco23
TL;DR: Refined descriptors that include consideration of disease activity (based on clinical relapse rate and imaging findings) and disease progression are proposed and strategies for future research to better define phenotypes are outlined.
Abstract: Accurate clinical course descriptions (phenotypes) of multiple sclerosis (MS) are important for communication, prognostication, design and recruitment of clinical trials, and treatment decision-making. Standardized descriptions published in 1996 based on a survey of international MS experts provided purely clinical phenotypes based on data and consensus at that time, but imaging and biological correlates were lacking. Increased understanding of MS and its pathology, coupled with general concern that the original descriptors may not adequately reflect more recently identified clinical aspects of the disease, prompted a re-examination of MS disease phenotypes by the International Advisory Committee on Clinical Trials of MS. While imaging and biological markers that might provide objective criteria for separating clinical phenotypes are lacking, we propose refined descriptors that include consideration of disease activity (based on clinical relapse rate and imaging findings) and disease progression. Strategies for future research to better define phenotypes are also outlined.
••
TL;DR: Detailed spatial statistical models of the channels are derived and it is found that, even in highly non-line-of-sight environments, strong signals can be detected 100-200 m from potential cell sites, potentially with multiple clusters to support spatial multiplexing.
Abstract: With the severe spectrum shortage in conventional cellular bands, millimeter wave (mmW) frequencies between 30 and 300 GHz have been attracting growing attention as a possible candidate for next-generation micro- and picocellular wireless networks. The mmW bands offer orders of magnitude greater spectrum than current cellular allocations and enable very high-dimensional antenna arrays for further gains via beamforming and spatial multiplexing. This paper uses recent real-world measurements at 28 and 73 GHz in New York, NY, USA, to derive detailed spatial statistical models of the channels and uses these models to provide a realistic assessment of mmW micro- and picocellular networks in a dense urban deployment. Statistical models are derived for key channel parameters, including the path loss, number of spatial clusters, angular dispersion, and outage. It is found that, even in highly non-line-of-sight environments, strong signals can be detected 100-200 m from potential cell sites, potentially with multiple clusters to support spatial multiplexing. Moreover, a system simulation based on the models predicts that mmW systems can offer an order of magnitude increase in capacity over current state-of-the-art 4G cellular networks with no increase in cell density from current urban deployments.
••
New York University1, Nathan Kline Institute for Psychiatric Research2, MIND Institute3, Katholieke Universiteit Leuven4, University of Utah5, Yale University6, University of California, Los Angeles7, Massachusetts Institute of Technology8, Trinity College, Dublin9, Carnegie Mellon University10, Ben-Gurion University of the Negev11, Ludwig Maximilian University of Munich12, Oregon Health & Science University13, Indiana University14, California Institute of Technology15, San Diego State University16, University of Groningen17, Netherlands Institute for Neuroscience18, University of Wisconsin-Madison19, Cornell University20, University of Pittsburgh21, Stanford University22, University of Michigan23, Kennedy Krieger Institute24, Johns Hopkins University25
TL;DR: W Whole-brain analyses reconciled seemingly disparate themes of both hypo- and hyperconnectivity in the ASD literature; both were detected, although hypoconnectivity dominated, particularly for corticocortical and interhemispheric functional connectivity.
Abstract: Autism spectrum disorders (ASDs) represent a formidable challenge for psychiatry and neuroscience because of their high prevalence, lifelong nature, complexity and substantial heterogeneity. Facing these obstacles requires large-scale multidisciplinary efforts. Although the field of genetics has pioneered data sharing for these reasons, neuroimaging had not kept pace. In response, we introduce the Autism Brain Imaging Data Exchange (ABIDE)-a grassroots consortium aggregating and openly sharing 1112 existing resting-state functional magnetic resonance imaging (R-fMRI) data sets with corresponding structural MRI and phenotypic information from 539 individuals with ASDs and 573 age-matched typical controls (TCs; 7-64 years) (http://fcon_1000.projects.nitrc.org/indi/abide/). Here, we present this resource and demonstrate its suitability for advancing knowledge of ASD neurobiology based on analyses of 360 male subjects with ASDs and 403 male age-matched TCs. We focused on whole-brain intrinsic functional connectivity and also survey a range of voxel-wise measures of intrinsic functional brain architecture. Whole-brain analyses reconciled seemingly disparate themes of both hypo- and hyperconnectivity in the ASD literature; both were detected, although hypoconnectivity dominated, particularly for corticocortical and interhemispheric functional connectivity. Exploratory analyses using an array of regional metrics of intrinsic brain function converged on common loci of dysfunction in ASDs (mid- and posterior insula and posterior cingulate cortex), and highlighted less commonly explored regions such as the thalamus. The survey of the ABIDE R-fMRI data sets provides unprecedented demonstrations of both replication and novel discovery. By pooling multiple international data sets, ABIDE is expected to accelerate the pace of discovery setting the stage for the next generation of ASD studies.
••
Harvard University1, Broad Institute2, Monash University3, Kyoto University4, Genentech5, Vanderbilt University6, New York University7, NewYork–Presbyterian Hospital8, Second Military Medical University9, University of Queensland10, University of Toronto11, University of Groningen12, University of Tartu13, Beijing Jiaotong University14, Icahn School of Medicine at Mount Sinai15, Radboud University Nijmegen16, Medisch Spectrum Twente17, Leiden University18, University of Paris19, French Institute of Health and Medical Research20, University of Alabama at Birmingham21, University of Amsterdam22, GlaxoSmithKline23, University of Cambridge24, Hanyang University25, Spanish National Research Council26, Complutense University of Madrid27, Umeå University28, Boston University29, Council on Education for Public Health30, McGill University31, University of Manchester32, National Health Service33, University of Pittsburgh34, University of California, San Francisco35, Karolinska Institutet36, North Shore-LIJ Health System37, University of Chicago38, University of Tokyo39
TL;DR: A genome-wide association study meta-analysis in a total of >100,000 subjects of European and Asian ancestries provides empirical evidence that the genetics of RA can provide important information for drug discovery, and sheds light on fundamental genes, pathways and cell types that contribute to RA pathogenesis.
Abstract: A major challenge in human genetics is to devise a systematic strategy to integrate disease-associated variants with diverse genomic and biological data sets to provide insight into disease pathogenesis and guide drug discovery for complex traits such as rheumatoid arthritis (RA)1. Here we performed a genome-wide association study meta-analysis in a total of >100,000 subjects of European and Asian ancestries (29,880 RA cases and 73,758 controls), by evaluating ~10 million single-nucleotide polymorphisms. We discovered 42 novel RA risk loci at a genome-wide level of significance, bringing the total to 101 (refs 2, 3, 4). We devised an in silico pipeline using established bioinformatics methods based on functional annotation5, cis-acting expression quantitative trait loci6 and pathway analyses7, 8, 9—as well as novel methods based on genetic overlap with human primary immunodeficiency, haematological cancer somatic mutations and knockout mouse phenotypes—to identify 98 biological candidate genes at these 101 risk loci. We demonstrate that these genes are the targets of approved therapies for RA, and further suggest that drugs approved for other indications may be repurposed for the treatment of RA. Together, this comprehensive genetic study sheds light on fundamental genes, pathways and cell types that contribute to RA pathogenesis, and provides empirical evidence that the genetics of RA can provide important information for drug discovery.
••
University of Bonn1, German Center for Neurodegenerative Diseases2, Harvard University3, Maastricht University4, Aix-Marseille University5, Pierre-and-Marie-Curie University6, French Institute of Health and Medical Research7, University of Melbourne8, VU University Amsterdam9, New York University10, Mayo Clinic11, City University of New York12, University of New South Wales13, Indiana University14, King's College London15, University of Toulouse16
TL;DR: Research criteria for SCD in pre‐mild cognitive impairment (MCI) are presented and a list of core features proposed for reporting in SCD studies is provided, which will enable comparability of research across different settings.
Abstract: There is increasing evidence that subjective cognitive decline (SCD) in individuals with unimpaired performance on cognitive tests may represent the first symptomatic manifestation of Alzheimer's disease (AD). The research on SCD in early AD, however, is limited by the absence of common standards. The working group of the Subjective Cognitive Decline Initiative (SCD-I) addressed this deficiency by reaching consensus on terminology and on a conceptual framework for research on SCD in AD. In this publication, research criteria for SCD in pre-mild cognitive impairment (MCI) are presented. In addition, a list of core features proposed for reporting in SCD studies is provided, which will enable comparability of research across different settings. Finally, a set of features is presented, which in accordance with current knowledge, increases the likelihood of the presence of preclinical AD in individuals with SCD. This list is referred to as SCD plus.
••
TL;DR: Bapineuzumab did not improve clinical outcomes in patients with Alzheimer's disease, despite treatment differences in biomarkers observed in APOE ε4 carriers.
Abstract: Background Bapineuzumab, a humanized anti-amyloid-beta monoclonal antibody, is in clinical development for the treatment of Alzheimer's disease. Methods We conducted two double-blind, randomized, placebo-controlled, phase 3 trials involving patients with mild-to-moderate Alzheimer's disease--one involving 1121 carriers of the apolipoprotein E (APOE) e4 allele and the other involving 1331 noncarriers. Bapineuzumab or placebo, with doses varying by study, was administered by intravenous infusion every 13 weeks for 78 weeks. The primary outcome measures were scores on the 11-item cognitive subscale of the Alzheimer's Disease Assessment Scale (ADAS-cog11, with scores ranging from 0 to 70 and higher scores indicating greater impairment) and the Disability Assessment for Dementia (DAD, with scores ranging from 0 to 100 and higher scores indicating less impairment). A total of 1090 carriers and 1114 noncarriers were included in the efficacy analysis. Secondary outcome measures included findings on positron-emission tomographic amyloid imaging with the use of Pittsburgh compound B (PIB-PET) and cerebrospinal fluid phosphorylated tau (phospho-tau) concentrations. Results There were no significant between-group differences in the primary outcomes. At week 78, the between-group differences in the change from baseline in the ADAS-cog11 and DAD scores (bapineuzumab group minus placebo group) were -0.2 (P=0.80) and -1.2 (P=0.34), respectively, in the carrier study; the corresponding differences in the noncarrier study were -0.3 (P=0.64) and 2.8 (P=0.07) with the 0.5-mg-per-kilogram dose of bapineuzumab and 0.4 (P=0.62) and 0.9 (P=0.55) with the 1.0-mg-per-kilogram dose. The major safety finding was amyloid-related imaging abnormalities with edema among patients receiving bapineuzumab, which increased with bapineuzumab dose and APOE e4 allele number and which led to discontinuation of the 2.0-mg-per-kilogram dose. Between-group differences were observed with respect to PIB-PET and cerebrospinal fluid phospho-tau concentrations in APOE e4 allele carriers but not in noncarriers. Conclusions Bapineuzumab did not improve clinical outcomes in patients with Alzheimer's disease, despite treatment differences in biomarkers observed in APOE e4 carriers. (Funded by Janssen Alzheimer Immunotherapy and Pfizer; Bapineuzumab 301 and 302 ClinicalTrials.gov numbers, NCT00575055 and NCT00574132, and EudraCT number, 2009-012748-17.).
••
TL;DR: Detailed information about the MNE package is given and typical use cases are described while also warning about potential caveats in analysis.
••
TL;DR: It is shown that low-dose penicillin (LDP), delivered from birth, induces metabolic alterations and affects ileal expression of genes involved in immunity, indicating that microbiota interactions in infancy may be critical determinants of long-term host metabolic effects.
•
08 Dec 2014TL;DR: In this paper, the authors exploit the redundancy present within the convolutional filters to derive approximations that significantly reduce the required computation, while keeping the accuracy within 1% of the original model.
Abstract: We present techniques for speeding up the test-time evaluation of large convolutional networks, designed for object recognition tasks. These models deliver impressive accuracy, but each image evaluation requires millions of floating point operations, making their deployment on smartphones and Internet-scale clusters problematic. The computation is dominated by the convolution operations in the lower layers of the model. We exploit the redundancy present within the convolutional filters to derive approximations that significantly reduce the required computation. Using large state-of-the-art models, we demonstrate speedups of convolutional layers on both CPU and GPU by a factor of 2 x, while keeping the accuracy within 1% of the original model.
•
TL;DR: This paper proposes a new hybrid architecture that consists of a deep Convolu-tional Network and a Markov Random Field and shows how this architecture is successfully applied to the challenging problem of articulated human pose estimation in monocular images.
Abstract: This paper proposes a new hybrid architecture that consists of a deep Convolutional Network and a Markov Random Field. We show how this architecture is successfully applied to the challenging problem of articulated human pose estimation in monocular images. The architecture can exploit structural domain constraints such as geometric relationships between body joint locations. We show that joint training of these two model paradigms improves performance and allows us to significantly outperform existing state-of-the-art techniques.
••
Christopher P. Ahn1, Rachael M. Alexandroff2, Carlos Allende Prieto3, Carlos Allende Prieto4 +272 more•Institutions (69)
TL;DR: The 10th public data release (DR10) from the Sloan Digital Sky Survey (SDSS-III) was released in 2013 as mentioned in this paper, which includes the first spectroscopic data from the Apache Point Observatory Galaxy Evolution Experiment (APOGEE), along with spectroscopy data from Baryon Oscillation Spectroscopic Survey (BOSS) taken through 2012 July.
Abstract: The Sloan Digital Sky Survey (SDSS) has been in operation since 2000 April. This paper presents the Tenth Public Data Release (DR10) from its current incarnation, SDSS-III. This data release includes the first spectroscopic data from the Apache Point Observatory Galaxy Evolution Experiment (APOGEE), along with spectroscopic data from the Baryon Oscillation Spectroscopic Survey (BOSS) taken through 2012 July. The APOGEE instrument is a near-infrared R ~ 22,500 300 fiber spectrograph covering 1.514-1.696 μm. The APOGEE survey is studying the chemical abundances and radial velocities of roughly 100,000 red giant star candidates in the bulge, bar, disk, and halo of the Milky Way. DR10 includes 178,397 spectra of 57,454 stars, each typically observed three or more times, from APOGEE. Derived quantities from these spectra (radial velocities, effective temperatures, surface gravities, and metallicities) are also included. DR10 also roughly doubles the number of BOSS spectra over those included in the Ninth Data Release. DR10 includes a total of 1,507,954 BOSS spectra comprising 927,844 galaxy spectra, 182,009 quasar spectra, and 159,327 stellar spectra selected over 6373.2 deg2.
••
Vanderbilt University1, Pacific Northwest National Laboratory2, Washington University in St. Louis3, Fred Hutchinson Cancer Research Center4, Icahn School of Medicine at Mount Sinai5, National Institutes of Health6, Massachusetts Institute of Technology7, Harvard University8, Georgetown University9, Johns Hopkins University10, Leidos11, Memorial Sloan Kettering Cancer Center12, National Institute of Standards and Technology13, New York University14, Stanford University15, University of Chicago16, University of North Carolina at Chapel Hill17, University of Washington18, Virginia Tech19
TL;DR: Integrated proteogenomic analysis provides functional context to interpret genomic abnormalities and affords a new paradigm for understanding cancer biology.
Abstract: Extensive genomic characterization of human cancers presents the problem of inference from genomic abnormalities to cancer phenotypes. To address this problem, we analysed proteomes of colon and rectal tumours characterized previously by The Cancer Genome Atlas (TCGA) and perform integrated proteogenomic analyses. Somatic variants displayed reduced protein abundance compared to germline variants. Messenger RNA transcript abundance did not reliably predict protein abundance differences between tumours. Proteomics identified five proteomic subtypes in the TCGA cohort, two of which overlapped with the TCGA 'microsatellite instability/CpG island methylation phenotype' transcriptomic subtype, but had distinct mutation, methylation and protein expression patterns associated with different clinical outcomes. Although copy number alterations showed strong cis- and trans-effects on mRNA abundance, relatively few of these extend to the protein level. Thus, proteomics data enabled prioritization of candidate driver genes. The chromosome 20q amplicon was associated with the largest global changes at both mRNA and protein levels; proteomics data highlighted potential 20q candidates, including HNF4A (hepatocyte nuclear factor 4, alpha), TOMM34 (translocase of outer mitochondrial membrane 34) and SRC (SRC proto-oncogene, non-receptor tyrosine kinase). Integrated proteogenomic analysis provides functional context to interpret genomic abnormalities and affords a new paradigm for understanding cancer biology.
••
University of Sydney1, University of California, Los Angeles2, Netherlands Cancer Institute3, New York University4, University of Groningen5, Women & Children's Hospital of Buffalo6, University of South Florida7, Royal Adelaide Hospital8, California Pacific Medical Center9, Princess Alexandra Hospital10, Roswell Park Cancer Institute11, Tom Baker Cancer Centre12
TL;DR: Biopsy-based staging of intermediate-thickness or thick primary melanomas provides important prognostic information and identifies patients with nodal metastases who may benefit from immediate complete lymphadenectomy, and prolongs disease-free survival and melanoma-specific survival.
Abstract: Background Sentinel-node biopsy, a minimally invasive procedure for regional melanoma staging, was evaluated in a phase 3 trial. Methods We evaluated outcomes in 2001 patients with primary cutaneous melanomas randomly assigned to undergo wide excision and nodal observation, with lymphadenectomy for nodal relapse (observation group), or wide excision and sentinel-node biopsy, with immediate lymphadenectomy for nodal metastases detected on biopsy (biopsy group). Results No significant treatment-related difference in the 10-year melanoma-specific survival rate was seen in the overall study population (20.8% with and 79.2% without nodal metastases). Mean (±SE) 10-year disease-free survival rates were significantly improved in the biopsy group, as compared with the observation group, among patients with intermediate-thickness melanomas, defined as 1.20 to 3.50 mm (71.3±1.8% vs. 64.7±2.3%; hazard ratio for recurrence or metastasis, 0.76; P=0.01), and those with thick melanomas, defined as >3.50 mm (50.7±4.0% vs...
••
University of Tokyo1, University of Wisconsin-Madison2, University of Toronto3, University of Kentucky4, University of Texas at Austin5, University of Washington6, University of St Andrews7, University of Portsmouth8, European Southern Observatory9, University of Nottingham10, National Autonomous University of Mexico11, University of Pittsburgh12, University of Cambridge13, New Mexico State University14, Carnegie Institution for Science15, University of Sydney16, New York University17, University of Utah18, University of Oxford19, University of California, Santa Cruz20, Max Planck Society21, Claude Bernard University Lyon 122, École normale supérieure de Lyon23, Texas Christian University24, University of Iowa25, Princeton University26, Case Western Reserve University27, University of La Laguna28, Chinese Academy of Sciences29, Academia Sinica30, University of Manchester31, Macquarie University32, Australian Astronomical Observatory33, Yale University34, Lawrence Berkeley National Laboratory35, University of Potsdam36, University of Victoria37, University of Groningen38
TL;DR: MaNGA (Mapping Nearby Galaxies at Apache Point Observatory) as mentioned in this paper employs dithered observations with 17 fiber-bundle integral field units that vary in diameter from 12'' (19 fibers) to 32'' (127 fibers).
Abstract: We present an overview of a new integral field spectroscopic survey called MaNGA (Mapping Nearby Galaxies at Apache Point Observatory), one of three core programs in the fourth-generation Sloan Digital Sky Survey (SDSS-IV) that began on 2014 July 1. MaNGA will investigate the internal kinematic structure and composition of gas and stars in an unprecedented sample of 10,000 nearby galaxies. We summarize essential characteristics of the instrument and survey design in the context of MaNGA's key science goals and present prototype observations to demonstrate MaNGA's scientific potential. MaNGA employs dithered observations with 17 fiber-bundle integral field units that vary in diameter from 12'' (19 fibers) to 32'' (127 fibers). Two dual-channel spectrographs provide simultaneous wavelength coverage over 3600-10300 A at R ~ 2000. With a typical integration time of 3 hr, MaNGA reaches a target r-band signal-to-noise ratio of 4-8 (A–1 per 2'' fiber) at 23 AB mag arcsec–2, which is typical for the outskirts of MaNGA galaxies. Targets are selected with M * 109 M ☉ using SDSS-I redshifts and i-band luminosity to achieve uniform radial coverage in terms of the effective radius, an approximately flat distribution in stellar mass, and a sample spanning a wide range of environments. Analysis of our prototype observations demonstrates MaNGA's ability to probe gas ionization, shed light on recent star formation and quenching, enable dynamical modeling, decompose constituent components, and map the composition of stellar populations. MaNGA's spatially resolved spectra will enable an unprecedented study of the astrophysics of nearby galaxies in the coming 6 yr.
••
St. Jude Children's Research Hospital1, University of Washington2, University of New Mexico3, National Institutes of Health4, Washington University in St. Louis5, University of Florida6, Ohio State University7, New York University8, University of Alabama at Birmingham9, University of Texas Southwestern Medical Center10, University of Pennsylvania11, Royal Children's Hospital12, University of Toronto13, Texas A&M University14, University of Utah15, Emory University16, Mayo Clinic17, University of Chicago18, University of Texas Health Science Center at San Antonio19, University of Texas MD Anderson Cancer Center20, Yeshiva University21, University of California, San Francisco22, University of Colorado Denver23
TL;DR: Ph-like ALL was found to be characterized by a range of genomic alterations that activate a limited number of signaling pathways, all of which may be amenable to inhibition with approved tyrosine kinase inhibitors.
Abstract: BACKGROUND Philadelphia chromosome–like acute lymphoblastic leukemia (Ph-like ALL) is characterized by a gene-expression profile similar to that of BCR–ABL1–positive ALL, alterations of lymphoid transcription factor genes, and a poor outcome. The frequency and spectrum of genetic alterations in Ph-like ALL and its responsiveness to tyrosine kinase inhibition are undefined, especially in adolescents and adults. METHODS We performed genomic profiling of 1725 patients with precursor B-cell ALL and detailed genomic analysis of 154 patients with Ph-like ALL. We examined the functional effects of fusion proteins and the efficacy of tyrosine kinase inhibitors in mouse pre-B cells and xenografts of human Ph-like ALL. RESULTS Ph-like ALL increased in frequency from 10% among children with standard-risk ALL to 27% among young adults with ALL and was associated with a poor outcome. Kinase-activating alterations were identified in 91% of patients with Ph-like ALL; rearrangements involving ABL1, ABL2, CRLF2, CSF1R, EPOR, JAK2, NTRK3, PDGFRB, PTK2B, TSLP, or TYK2 and sequence mutations involving FLT3, IL7R, or SH2B3 were most common. Expression of ABL1, ABL2, CSF1R, JAK2, and PDGFRB fusions resulted in cytokine-independent proliferation and activation of phosphorylated STAT5. Cell lines and human leukemic cells expressing ABL1, ABL2, CSF1R, and PDGFRB fusions were sensitive in vitro to dasatinib, EPOR and JAK2 rearrangements were sensitive to ruxolitinib, and the ETV6–NTRK3 fusion was sensitive to crizotinib. CONCLUSIONS Ph-like ALL was found to be characterized by a range of genomic alterations that activate a limited number of signaling pathways, all of which may be amenable to inhibition with approved tyrosine kinase inhibitors. Trials identifying Ph-like ALL are needed to assess whether adding tyrosine kinase inhibitors to current therapy will improve the survival of patients with this type of leukemia. (Funded by the American Lebanese Syrian Associated Charities and others.)
••
Columbia University1, University of Pennsylvania2, Rush University Medical Center3, University of Kentucky4, Uppsala University5, Newcastle University6, Northwestern University7, Washington University in St. Louis8, Mayo Clinic9, Emory University10, University of São Paulo11, University of California12, Icahn School of Medicine at Mount Sinai13, Harvard University14, Medical University of Vienna15, University of Pittsburgh16, University of Washington17, University of British Columbia18, University of California, San Diego19, Boston University20, University of California, San Francisco21, University of Iowa22, University of Ulm23, University of Texas Southwestern Medical Center24, New York University25, Oregon Health & Science University26, Kanazawa University27
TL;DR: A new term is recommended, “primary age-related tauopathy” (PART), to describe a pathology that is commonly observed in the brains of aged individuals, yet this pathological process cannot be specifically identified pre-mortem at the present time.
Abstract: We recommend a new term, "primary age-related tauopathy" (PART), to describe a pathology that is commonly observed in the brains of aged individuals. Many autopsy studies have reported brains with neurofibrillary tangles (NFTs) that are indistinguishable from those of Alzheimer's disease (AD), in the absence of amyloid (Aβ) plaques. For these "NFT+/Aβ-" brains, for which formal criteria for AD neuropathologic changes are not met, the NFTs are mostly restricted to structures in the medial temporal lobe, basal forebrain, brainstem, and olfactory areas (bulb and cortex). Symptoms in persons with PART usually range from normal to amnestic cognitive changes, with only a minority exhibiting profound impairment. Because cognitive impairment is often mild, existing clinicopathologic designations, such as "tangle-only dementia" and "tangle-predominant senile dementia", are imprecise and not appropriate for most subjects. PART is almost universally detectable at autopsy among elderly individuals, yet this pathological process cannot be specifically identified pre-mortem at the present time. Improved biomarkers and tau imaging may enable diagnosis of PART in clinical settings in the future. Indeed, recent studies have identified a common biomarker profile consisting of temporal lobe atrophy and tauopathy without evidence of Aβ accumulation. For both researchers and clinicians, a revised nomenclature will raise awareness of this extremely common pathologic change while providing a conceptual foundation for future studies. Prior reports that have elucidated features of the pathologic entity we refer to as PART are discussed, and working neuropathological diagnostic criteria are proposed.
••
TL;DR: In this paper, the authors show that initial reports cause high-frequency "action and backsliding", but these cycles attenuate over time. And if reports are discontinued after two years, effects are relatively persistent, decaying at 10-20 percent per year.
Abstract: We document three remarkable features of the Opower program, in which social comparison-based home energy reports are repeatedly mailed to more than six million households nationwide. First, initial reports cause high-frequency “action and backsliding,” but these cycles attenuate over time. Second, if reports are discontinued after two years, effects are relatively persistent, decaying at 10–20 percent per year. Third, consumers are slow to habituate: they continue to respond to repeated treatment even after two years. We show that the previous conservative assumptions about post-intervention persistence had dramatically understated cost effectiveness and illustrate how empirical estimates can optimize program design.(JEL D12, D83, L94, Q41)