检索结果-内蒙古大学图书馆

IEEE/ACM Workshop on machine learning in High Performance computing Environments (MLHPC)

作者： Avraam Chatzimichailidis Janis Keuper Franz-Josef Pfreundt Nicolas R. Gauger Competence Center High Performance Computing Fraunhofer ITWM Kaiserslautern Germany Chair for Scientific Computing TU Kaiserslautern Kaiserslautern Germany Institute for Machine Learning and Analytics Offenburg University Germany Fraunhofer Center Machine Learning Germany

ISBN: (纸本)9781728159867

Current training methods for deep neural networks boil down to very high dimensional and non-convex optimization problems which are usually solved by a wide range of stochastic gradient descent methods. While these approaches tend to work in practice, there are still many gaps in the theoretical understanding of key aspects like convergence and generalization guarantees, which are induced by the properties of the optimization surface (loss landscape). In order to gain deeper insights, a number of recent publications proposed methods to visualize and analyze the otimization surfaces. However, the computational cost of these methods are very high, making it hardly possible to use them on larger networks. In this paper, we present the GradVis Toolbox, an open source library for efficient and scalable visualization and analysis of deep neural network loss landscapes in Tesorflow and PyTorch. Introducing more efficient mathematical formulations and a novel parallelization scheme, GradVis allows to plot 2d and 3d projections of optimization surfaces and trajectories, as well as high resolution second order gradient information for large networks.

关键词： Visualization Eigenvalues Parallelization Deep-learning Second-order parallelization Visualization eigenvalues Neural network source program library Training Functional training Surfaces Analysis

来源：评论

学校读者我要写书评

暂无评论

Scalable Hyperparameter Optimization with Lazy Gaussian Processes

Scalable Hyperparameter Optimization with Lazy Gaussian Proc...

引用

IEEE/ACM Workshop on machine learning in High Performance computing Environments (MLHPC)

作者： Raju Ram Sabine Müller Franz-Josef Pfreundt Nicolas R. Gauger Janis Keuper Competence Center High Performance Computing Fraunhofer ITWM Kaiserslautern Germany Scientific Computing Group TU Kaiserslautern Germany Fraunhofer ITWM Kaiserslautern Germany Fraunhofer Center Machine Learning Germany Institute for Machine Learning and Analytics Offenburg University Germany

ISBN: (纸本)9781728159867

Most machine learning methods require careful selection of hyper-parameters in order to train a high performing model with good generalization abilities. Hence, several automatic selection algorithms have been introduced to overcome tedious manual (try and error) tuning of these parameters. Due to its very high sample efficiency, Bayesian Optimization over a Gaussian Processes modeling of the parameter space has become the method of choice. Unfortunately, this approach suffers from a cubic compute complexity due to underlying Cholesky factorization, which makes it very hard to be scaled beyond a small number of sampling steps. In this paper, we present a novel, highly accurate approximation of the underlying Gaussian Process. Reducing its computational complexity from cubic to quadratic allows an efficient strong scaling of Bayesian Optimization while outperforming the previous approach regarding optimization accuracy. First experiments show speedups of a factor of 162 in single node and further speed up by a factor of 5 in a parallel environment.

关键词： Hyper-Parameter optimization Neural architecture search Bayesian optimization Gaussian processes Scalability Gaussian processes Scalability Hyperparameter parameter space methods Cholesky factorisation Cubic Speeding

来源：评论

学校读者我要写书评

暂无评论

A multimodal digital biomarker of functional deficits in early-stage Alzheimer’s disease: results of the RADAR-AD study

引用

Alzheimer's & Dementia 2023年第S15期19卷

作者： Srinivasan Vairavan Manuel Lentzen Marijn Muurling Casper de Boer Alankar Atreya Jelena Curcic Chris Hinds Pauline Conde Margarita Grammatikopoulou Gaetano Scebba Ioulietta Lazarou Spiros Nikolopoulos Anna-Katharine Brem Neva Coello Pieter Jelle Visser Holger Fröhlich Vaibhav A Narayan Gayle Wittenberg Dag Aarsland Janssen Research and Development Titusville NJ USA Correspondece Fraunhofer Institute for Algorithms and Scientific Computing SCAI Sankt Augustin Germany VU University Medical Center Amsterdam Netherlands University of Oxford Oxford United Kingdom Novartis Institutes for BioMedical Research Basel Switzerland King’s College London London United Kingdom Centre for Research & Technology Hellas Thessaloniki Greece Davos Alzheimers Collaborative Wayne PA USA Janssen Research and Development LLC Titusville NJ USA

Background Remote monitoring technologies (RMTs), such as smartphone apps, smartwatches, and in-home sensors, are rapidly changing the way functional and cognitive performance is measured in Alzheimer’s disease (AD) patients. Here, we present results from the European RADAR-AD study on the use of multimodal data streams for the identification of functional deficits across all syndromic stages of AD. Method Four study groups (Healthy controls (HC), preclinical AD (pre. AD), prodromal AD (pro. AD), and mild AD) were included in this cross-sectional study. The RMT features (gait measures from Timed up and Go (TUG), Dual Task Effect (DTE) using physilog, acoustic features from Speech task in Mezurio, neurocognitive function using Altoida, managing finances with Banking app) with in-clinic neuropsychological (NP) tests, activities of daily living with Amsterdam IADL and demographics (age, gender, education years) were analyzed for different combinations of the multimodal digital biomarker of disease stage in AD across different pairwise comparisons. The analysis includes data from 175 participants (HC = 67, *** = 26, *** = 50, Mild AD = 32) collected for 8 weeks. An extreme gradient boosting (XGBoost) machine learning model was trained to obtain a multimodal biomarker of AD disease stage with repeated cross-validation (5-fold with 4 repeats) (Figure 1). This investigation is part of the ongoing RADAR-AD study. Result The multimodal combination of RMTs achieved a mean AUC of > 0.60 in all pairwise comparisons with a mean AUC > 0.65 for HC vs (***, *** and Mild) (Figures 2 and 3). The addition of NP tests increases the performance considerably across all pairwise comparisons except HC vs ***. Conclusion Our results highlight the advantage of combining RMTs to identify functional deficits in the early stage of AD. In particular, in prodromal and mild AD patients, a combined signal shows much more strength compared to individual tests. This work has received s

关键词：

来源：评论

学校读者我要写书评

暂无评论

Redefining pandemic preparedness:Multidisciplinary insights from the CERP modelling workshop in infectious diseases,workshop report

引用

Infectious Disease Modelling 2024年第2期9卷 501-518页

作者： Marta C.Nunes Edward Thommes Holger Fröhlich Antoine Flahault Julien Arino Marc Baguelin Matthew Biggerstaff Gaston Bizel-Bizellot Rebecca Borchering Giacomo Cacciapaglia Simon Cauchemez Alex Barbier-Chebbah Carsten Claussen Christine Choirat Monica Cojocaru Catherine Commaille-Chapus Chitin Hon Jude Kong Nicolas Lambert Katharina B.Lauer Thorsten Lehr Cédric Mahe Vincent Marechal Adel Mebarki Seyed Moghadas Rene Niehus Lulla Opatowski Francesco Parino Gery Pruvost Andreas Schuppert Rodolphe Thiébaut Andrea Thomas-Bachli Cecile Viboud Jianhong Wu Pascal Crépey Laurent Coudeville Center of Excellence in Respiratory Pathogens(CERP) Hospices Civils de Lyon(HCL)and Centre International de Recherche en Infectiologie(CIRI)Équipe SantéPubliqueÉpidémiologie etÉcologieÉvolutive des Maladies Infectieuses(PHE3ID)Inserm U1111CNRS UMR5308ENS de LyonUniversitéClaude Bernard Lyon 1LyonFrance South African Medical Research Council Vaccines&Infectious Diseases Analytics Research UnitFaculty of Health SciencesUniversity of the WitwatersrandJohannesburgSouth Africa New Products and Innovation(NPI) Sanofi Vaccines(Global)TorontoOntarioCanada Department of Mathematics and Statistics University of GuelphGuelphOntarioCanada Fraunhofer Institute for Algorithms and Scientific Computing(SCAI) Department of BioinformaticsSchloss BirlinghovenSankt AugustinGermany University of Bonn Bonn-Aachen International Center for IT(b-it)BonnGermany Institute of Global Health Faculty of MedicineUniversity of GenevaGenevaSwitzerland and Swiss School of Public HealthZürichSwitzerland Department of Mathematics University of ManitobaWinnipegManitobaCanada MRC Centre for Global Infectious Disease Analysis School of Public HealthImperial College LondonLondonUK Centre for Mathematical Modelling of Infectious Diseases Department of Infectious Disease EpidemiologyLondon School of Hygiene&Tropical MedicineLondonUK National Center for Immunization and Respiratory Diseases(NCIRD) US Centers for Disease Control and Prevention(CDC)AtlantaGAUSA Departement of Computational Biology Departement of Global HealthInstitut PasteurParisFrance Institut de Physique des Deux Infinis de Lyon(IP2I) UMR5822IN2P3/CNRSUniversitéClaude Bernard Lyon 1VilleurbanneFrance Mathematical Modelling of Infectious Diseases Unit Institut PasteurUniversitéParis CitéUMR2000 CNRSParisFrance Decision and Bayesian Computation Institut PasteurUniversitéParis CitéCNRS UMR 3571France Fraunhofer-Institute for Translational Medicine and Pharmacology HamburgGermany Institute of Global Health Faculty of MedicineUniversity of

In July 2023,the center of Excellence in Respiratory Pathogens organized a two-day workshop on infectious diseases modelling and the lessons learnt from the Covid-19 *** report summarizes the rich discussions that occurred during the *** workshop participants discussed multisource data integration and highlighted the benefits of combining traditional surveillance with more novel data sources like mobility data,social media,and wastewater *** advancements were noted in the development of predictive models,with examples from various countries showcasing the use of machine learning and artificial intelligence in detecting and monitoring disease *** role of open collaboration between various stakeholders in modelling was stressed,advocating for the continuation of such partnerships beyond the pandemic.A major gap identified was the absence of a common international framework for data sharing,which is crucial for global pandemic ***,the workshop underscored the need for robust,adaptable modelling frameworks and the integration of different data sources and collaboration across sectors,as key elements in enhancing future pandemic response and preparedness.

关键词： Modelling Covid-19 Infectious diseases Pandemic preparedness Workshop

来源：评论

学校读者我要写书评

暂无评论

Progression subtypes in Parkinson's disease: Results of an AI-based multi-cohort analysis

引用

Clinical Neurophysiology 2024年 159卷 e39-e39页

作者： Hähnel, T. Raschka, T. Sapienza, S. Klucken, J. Glaab, E. Corvol, J. Falkenburger, B. Fröhlich, H. Department of Neurology University Hospital and Faculty of Medicine Carl Gustav Carus TUD Dresden University of Technology Dresden Germany Department of Bioinformatics Fraunhofer Institute for Algorithms and Scientific Computing (SCAI) Sankt Augustin Germany Bonn-Aachen International Center for IT University of Bonn Bonn Germany Biomedical Data Science Luxembourg Centre for Systems Biomedicine (LCSB) University of Luxembourg Esch-sur-Alzette Luxembourg Luxembourg Institute of Health (LIH) Strassen Luxembourg Centre Hospitalier de Luxembourg (CHL) Luxembourg Sorbonne Université Paris Brain Institute – ICM Inserm CNRS Assistance Publique Hôpitaux de Paris Pitié-Salpêtrière Hospital Department of Neurology Paris France German Center for Neurodegenerative Diseases (DZNE) Tatzberg 41 01307 Dresden Germany

来源：评论

学校读者我要写书评

暂无评论

Assessment of RMTs for Discriminating Stages of Alzheimer’s Disease

引用

Alzheimer's & Dementia 2023年第S22期19卷

作者： Manuel Lentzen Srinivasan Vairavan Jelena Curcic Alankar Atreya Chris Hinds Marijn Muurling Pauline Conde Margarita Grammatikopoulou Ioulietta Lazarou Spiros Nikolopoulos Neva Coello Casper de Boer Dag Aarsland Anna-Katharine Brem Holger Fröhlich RADAR-AD Fraunhofer Institute for Algorithms and Scientific Computing SCAI Sankt Augustin Germany Bonn-Aachen International Center for IT (b-it) Bonn Germany Janssen Research and Development LLC Titusville NJ USA Novartis Institutes for BioMedical Research Basel Switzerland University of Oxford Oxford United Kingdom VU University Medical Center Amsterdam Netherlands Alzheimer Center Amsterdam Neurology Vrije Universiteit Amsterdam Amsterdam UMC location VUmc Amsterdam Netherlands Amsterdam Neuroscience Neurodegeneration Amsterdam Netherlands King’s College London London United Kingdom Centre for Research & Technology Hellas Thessaloniki Greece Amsterdam Neuroscience Neurodegeneration Amsterdam Noord-Holland Netherlands Alzheimer Center Amsterdam Neurology Vrije Universiteit Amsterdam Amsterdam UMC location VUmc Amsterdam Noord-Holland Netherlands

Background RADAR-AD is a European project in the context of the Innovative Medicine Initiative (IMI) focusing on the earlier identification of patients at risk for developing Alzheimer’s Disease (AD) via a panel of remote monitoring technologies (RMTs), including smartphone apps and wearable devices. Method We examined the ability of 6 RMTs (Altoida, Axivity, Banking app, Fitbit, Physilog, and Mezurio) to distinguish between healthy controls (HC) and disease stages of preclinical (PreAD), prodromal (ProAD), and mild to moderate Alzheimer’s disease (MildAD) based on 175 patients (interim analysis). We trained three machine learning classifiers (Logistic Regression, Random Forest, and XGBoost) in a pairwise setting (HC vs. PreAD, HC vs. ProAD, HC vs. MildAD, PreAD vs. ProAD, and ProAD vs. MildAD). Since the interim dataset is still limited, we performed repeated, stratified nested cross-validation to get a robust performance estimate. Each classifier was trained with the features of the different devices and a set of baseline variables. The latter include a patient’s gender, age, years of education, and body mass index (BMI) when physical conditions might play a role (Axivitiy, Fitbit, Physilog). In addition, we checked whether specific patterns of the study groups allowed discrimination of the different study groups based on the baseline variables alone. Therefore, we trained one Logistic Regression model with these variables and compared the performance of the other three models with this baseline. The models trained with the baseline and questionnaire-based data served as the reference value in our benchmark that represents how well the discrimination of the different groups works with clinical tests. Result Our preliminary data show that RMTs can identify patients already in a prodromal disease stage (AUC ∼69%, Figure 1). Furthermore, the pairwise combination of data from a banking app and an app monitoring functional cognitive abilities via an augmented reality g

关键词：

来源：评论

学校读者我要写书评

暂无评论

Evaluating digital device technology in Alzheimer’s disease via artificial intelligence

引用

Alzheimer's & Dementia 2021年第S6期17卷

作者： Meemansa Sood Robbert Harms Maximilian Bügler Ioannis Tarnanas Neva Coello Holger Fröhlich Bonn-Aachen International Center for IT (b-it) Bonn Germany Fraunhofer Institute for Algorithms and Scientific Computing SCAI Sankt Augustin Germany Altoida Inc. Houston TX USA Altoida AG Hirschengraben Lucerne Switzerland Global Brain Health Institute Trinity College Dublin Ireland Hellenic Initiative Against Alzheimer's Disease Johns Hopkins Precision Medicine Center Baltimore MD USA Novartis Basel Switzerland

Background Digital Measures (DMs) derived from mobile devices and smartphone applications have received a strongly increasing attention during the last years, because they could allow for an accurate, quantitative monitoring of disease symptoms, even outside clinics. In addition, DMs may help to diagnose Alzheimer’s Disease (AD) in a pre-symptomatic stage and thus increase the success chances of therapeutic interventions. However, before any use in clinical routine, DMs have to be evaluated carefully by assessing their relationship to established clinical scores and understanding their diagnostic benefit. In this regard the IMI project RADAR-AD ( *** ) has the ambition to evaluate a broad panel of digital technologies with respect to their potential for early disease diagnosis while focusing on functional activities of daily living. Method An example of a panel of digital technology RADAR-AD uses is a smartphone based virtual reality game resulting into an assessment of cognitive impairment. In our work we analyzed connections between digital readouts and cognitive features like MMSE (Mini Mental State Examination) via one of our recently developed Artificial Intelligence (AI) approaches called Variational Autoencoder Modular Bayesian Networks (VAMBN). Going one step further we also tested the possibility to accurately predict MMSE scores from DMs and vice versa via machine learning. Based on this finding we then simulated DMs within the ADNI cohort and re-ran VAMBN. Result Application of VAMBN on the data from virtual reality game resulted into a network comprising DMs, MMSE sub-item scores and demographic features (Figure 1). It thus allowed to disentangle and quantify the relationship between DMs and established clinical scores. The simulation of DM’s and application of VAMBN in the ADNI cohort allowed us to further predict connections of DMs with FAQ (Functional Activity Questionnaire) and even molecular mechanisms. Conclusion Our results indicate t

关键词：

来源：评论

学校读者我要写书评

暂无评论

A representer theorem for deep kernel learning

The Journal of Machine Learning Research

引用

The Journal of machine learning Research 2019年第1期20卷

作者： Bastian Bohn Christian Rieger Michael Griebel Institute for Numerical Simulation University of Bonn Bonn Germany Institute for Numerical Simulation University of Bonn Bonn Germany and Fraunhofer Center for Machine Learning Fraunhofer Institute for Algorithms and Scientific Computing SCAI Sankt Augustin Germany

In this paper we provide a finite-sample and an infinite-sample representer theorem for the concatenation of (linear combinations of) kernel functions of reproducing kernel Hilbert spaces. These results serve as mathematical foundation for the analysis of machine learning algorithms based on compositions of functions. As a direct consequence in the finite-sample case, the corresponding infinite-dimensional minimization problems can be recast into (nonlinear) finite-dimensional minimization problems, which can be tackled with nonlinear optimization algorithms. Moreover, we show how concatenated machine learning problems can be reformulated as neural networks and how our representer theorem applies to a broad class of state-of-the-art deep learning methods.

关键词： artificial neural networks deep kernel learning multilayer kernel regularized least-squares regression representer theorem

来源：评论

学校读者我要写书评

暂无评论

Motor-cognitive dual tasking in the clinical setting: a sensitive measure of functional impairment in early Alzheimer’s disease

引用

Alzheimer's & Dementia 2023年第S14期19卷

作者： Anna-Katharine Brem Gaetano Scebba Jelena Curcic Marijn Muurling Casper de Boer Neva Coello Alankar Atreya Pauline Conde Holger Fröhlich Margarita Grammatikopoulou Chris Hinds Ioulietta Lazarou Manuel Lentzen Vaibhav A Narayan Rouba Kozak Spiros Nikolopoulos Srinivasan Vairavan Pieter Jelle Visser Gayle Wittenberg Dag Aarsland RADAR-AD Consortium University Hospital of Old Age Psychiatry University of Bern Bern Switzerland King’s College London London United Kingdom Correspondece Novartis Institutes for BioMedical Research Basel Switzerland Alzheimer Center Amsterdam Neurology Vrije Universiteit Amsterdam Amsterdam UMC location VUmc Amsterdam Netherlands Alzheimer Center Amsterdam Neurology Vrije Universiteit Amsterdam Amsterdam UMC location VUmc Amsterdam Noord-Holland Netherlands University of Oxford Oxford United Kingdom Fraunhofer Institute for Algorithms and Scientific Computing SCAI Sankt Augustin Germany Centre for Research & Technology Hellas Thessaloniki Greece Davos Alzheimers Collaborative Wayne PA USA Novartis Cambridge MA USA Janssen Research and Development LLC Titusville NJ USA Alzheimer Center Amsterdam Department of Neurology Amsterdam University Medical Center Amsterdam Netherlands Maastricht University Maastricht Netherlands Stavanger University Hospital Stavanger Norway

Background Gait is a complex everyday activity that depends upon supraspinal activity and a host of cognitive functions such as attention and executive functions. As cognition declines in neurodegenerative diseases, the interaction and competition for neuronal resources during motor-cognitive dual-tasking (e.g., walking while talking) might be a sensitive measure of subtle functional impairments in early Alzheimer’s disease (AD). Here, we aim to identify gait deficits due to neuronal competition across the AD spectrum. Method This investigation is part of the ongoing Remote Assessment of Disease and Relapse – Alzheimer’s Disease (RADAR-AD) study. We attached three inertial measurement units (accelerometer and gyroscope) to both feet and one hip to assess dual task effects (DTE) assessing gait performance with/without concurrent serial subtraction-by-1 task in four groups: 1) amyloid negative healthy controls (HC, N = 59); and 2) amyloid positive preclinical AD (PreAD, N = 30); 3) prodromal AD (ProAD, N = 51); and 4) mild-to-moderate AD dementia (MildAD, N = 44) (Table 1). We furthermore investigated associations of DTE with observer-reported cognition. Result Group comparisons showed that dual-tasking induced lower cadence and increased stance, which were significantly different between HC and ProAD. Several DTE measures of variability differed significantly between PreAD and MildAD, with variability in the path length separating best between PreAD and ProAD (Table 2, Figure 1). DTE measures were associated with observer-rated divided attention only in the MildAD group. Conclusion Neuronal competition as assessed with motor-cognitive dual-tasking, specifically the DTE variability, might reflect functional deficits already in early AD, and could be a valuable additional measure to detect early impairments not captured by cognitive or motor tests alone. Future studies should implement an adaptive cognitive load to improve sensitivity/specificity in early AD stages and

关键词：

来源：评论

学校读者我要写书评

暂无评论

NightVision: Generating nighttime satellite imagery from infra-red observations

arXiv

引用

arXiv 2020年

作者： Harder, Paula Jones, William Lguensat, Redouane Bouabid, Shahine Fulton, James Marcolongo, Aris Rao, Yuhan Quesada-Chacón, Dánnell Stefanović, Sofija Manshausen, Peter Watson-Parris, Duncan Fraunhofer Center Machine Learning Scientific Computing University of Kaiserlautern Germany Dpt. of Atmospheric Oceanic and Planetary Physics University of Oxford United Kingdom LSCE-IPSL CEA Saclay LOCEAN-IPSL Sorbonne Université France Dpt. of Statistics University of Oxford United Kingdom School of Geosciences University of Edinburgh United Kingdom Mathematical Institute University of Bern Climate and Environmental Physics University of Bern Switzerland Institute for Climate Studies North Carolina State University United States Institute of Hydrology and Meteorology Dresden University of Technology Germany

The recent explosion in applications of machine learning to satellite imagery often rely on visible images and therefore suffer from a lack of data during the night. The gap can be filled by employing available infra-red observations to generate visible images. This work presents how deep learning can be applied successfully to create those images by using U-Net based architectures. The proposed methods show promising results, achieving a structural similarity index (SSIM) up to 86% on an independent test set and providing visually convincing output images, generated from infra-red observations. The code is available at: https://***/paulaharder/hackathon-ci-2020 Copyright © 2020, The Authors. All rights reserved.

关键词： Satellite imagery

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：