检索结果-内蒙古大学图书馆

arXiv 2019年

作者： Gupta, Chirag Balakrishnan, Sivaraman Ramdas, Aaditya Machine Learning Department Carnegie Mellon University United States Department of Statistics and Data Science Carnegie Mellon University United States

We derive bounds on the path length ζof gradient descent (GD) and gradient flow (GF) curves for various classes of smooth convex and nonconvex functions. Among other results, we prove that: (a) if the iterates are linearly convergent with factor(1â'c), then ζ is at mostO(1/c);(b) under the Polyak-Kurdyka-Lojasiewicz (PKL) condition,ζis at most O(κâ'â'√), whereκis the condition number, and at least Ω(dâ'â'√∧κ1/4);(c) for quadratics,ζis Θ(min{dâ'â'√,logκâ'â'â'â'√})and in some cases can be independent ofκ;(d) assuming just convexity,ζcan be at most 24dlogd;(e) for separable quasiconvex functions,ζis Θ(dâ'â'√). Thus, we advance current understanding of the properties of GD and GF curves beyond rates of convergence. We expect our techniques to facilitate future studies for other algorithms. Copyright © 2019, The Authors. All rights reserved.

关键词： Gradient methods

来源：评论

学校读者我要写书评

暂无评论

Cosmological N-body simulations: a challenge for scalable generative models

引用

Computational Astrophysics and Cosmology 2019年第1期6卷 1-17页

作者： Perraudin, Nathanaël Srivastava, Ankit Lucchi, Aurelien Kacprzak, Tomasz Hofmann, Thomas Réfrégier, Alexandre Swiss Data Science Center ETH Zurich Zurich Switzerland Institute for Machine Learning ETH Zurich Zurich Switzerland Institute for Particle Physics and Astrophysics ETH Zurich Zurich Switzerland

Deep generative models, such as Generative Adversarial Networks (GANs) or Variational Autoencoders (VAs) have been demonstrated to produce images of high visual quality. However, the existing hardware on which these models are trained severely limits the size of the images that can be generated. The rapid growth of high dimensional data in many fields of science therefore poses a significant challenge for generative models. In cosmology, the large-scale, three-dimensional matter distribution, modeled with N-body simulations, plays a crucial role in understanding the evolution of structures in the universe. As these simulations are computationally very expensive, GANs have recently generated interest as a possible method to emulate these datasets, but they have been, so far, mostly limited to two dimensional data. In this work, we introduce a new benchmark for the generation of three dimensional N-body simulations, in order to stimulate new ideas in the machine learning community and move closer to the practical use of generative models in cosmology. As a first benchmark result, we propose a scalable GAN approach for training a generator of N-body three-dimensional cubes. Our technique relies on two key building blocks, (i) splitting the generation of the high-dimensional data into smaller parts, and (ii) using a multi-scale approach that efficiently captures global image features that might otherwise be lost in the splitting process. We evaluate the performance of our model for the generation of N-body samples using various statistical measures commonly used in cosmology. Our results show that the proposed model produces samples of high visual quality, although the statistical analysis reveals that capturing rare features in the data poses significant problems for the generative models. We make the data, quality evaluation routines, and the proposed GAN architecture publicly available at https://***/nperraud/3DcosmoGAN.

关键词：

来源：评论

学校读者我要写书评

暂无评论

The medical algorithmic audit (vol 4, pg e384, 2022)

引用

LANCET DIGITAL HEALTH 2022年第6期4卷 E405-E405页

作者： Liu, X. Glocker, B. McCradden, M. M. Ghassemi, M. Denniston, A. K. Oakden-Rayner, L. Academic Unit of Ophthalmology Institute of Inflammation and Ageing College of Medical and Dental Sciences University of Birmingham UK Department of Ophthalmology University Hospitals Birmingham NHS Foundation Trust Birmingham UK Moorfields Eye Hospital NHS Foundation Trust London UK Health Data Research UK London UK Birmingham Health Partners Centre for Regulatory Science and Innovation University of Birmingham Birmingham UK Biomedical Image Analysis Group Department of Computing Imperial College London London UK The Hospital for Sick Children Toronto ON Canada Dalla Lana School of Public Health Toronto ON Canada Institute for Medical Engineering and Science and Department of Electrical Engineering and Computer Science Massachusetts Institute of Technology Cambridge MA USA National Institute of Health Research Biomedical Research Centre for Ophthalmology Moorfields Hospital London NHS Foundation Trust London UK University College London Institute of Ophthalmology London UK Australian Institute for Machine Learning University of Adelaide Adelaide SA Australia. lauren.oakden-rayner@adelaide.edu.au

Artificial intelligence systems for health care, like any other medical device, have the potential to fail. However, specific qualities of artificial intelligence systems, such as the tendency to learn spurious correlates in training data, poor generalisability to new deployment settings, and a paucity of reliable explainability mechanisms, mean they can yield unpredictable errors that might be entirely missed without proactive investigation. We propose a medical algorithmic audit framework that guides the auditor through a process of considering potential algorithmic errors in the context of a clinical task, mapping the components that might contribute to the occurrence of errors, and anticipating their potential consequences. We suggest several approaches for testing algorithmic errors, including exploratory error analysis, subgroup testing, and adversarial testing, and provide examples from our own work and previous studies. The medical algorithmic audit is a tool that can be used to better understand the weaknesses of an artificial intelligence system and put in place mechanisms to mitigate their impact. We propose that safety monitoring and medical algorithmic auditing should be a joint responsibility between users and developers, and encourage the use of feedback mechanisms between these groups to promote learning and maintain safe deployment of artificial intelligence *** © 2022 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 license. Published by Elsevier Ltd.. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Sharpening the dark matter signature in gravitational waveforms II: Numerical simulations with the NbodyIMRI code

arXiv

引用

arXiv 2024年

作者： Kavanagh, Bradley J. Karydas, Theophanes K. Bertone, Gianfranco Di Cintio, Pierfrancesco Pasquato, Mario Av. de Los Castros s/n Santander39005 Spain Institute for Theoretical Physics Amsterdam and Delta Institute for Theoretical Physics University of Amsterdam Science Park 904 Amsterdam1098 XH Netherlands 50022 Italy INAF-Osservatorio Astronomico di Arcetri Largo Enrico Fermi 5 Firenze50125 Italy INFN-Sezione di Firenze via G. Sansone 1 Sesto Fiorentino50022 Italy Département de Physique Université de Montréal 1375 Avenue Thérèse-Lavoie-Roux Montréal Canada Mila – Quebec Artificial Intelligence Institute 6666 Rue Saint-Urbain Montréal Canada Ciela – Montréal Institute for Astrophysical Data Analysis and Machine Learning Montréal Canada Dipartimento di Fisica e Astronomia Università di Padova Vicolo dell’Osservatorio 5 Padova Italy Istituto Nazionale di Fisica Nucleare Padova Via Marzolo 8 Padova Italy

Future gravitational wave observatories can probe dark matter by detecting the dephasing in the waveform of binary black hole mergers induced by dark matter overdensities. Such a detection hinges on the accurate modelling of the dynamical friction, induced by dark matter on the secondary compact object in intermediate and extreme mass ratio inspirals. In this paper, we introduce NbodyIMRI, a new publicly available code designed for simulating binary systems within cold dark matter ‘spikes’. Leveraging higher particle counts and finer timesteps, we validate the applicability of the standard dynamical friction formalism and provide an accurate determination of the maximum impact parameter of particles which can effectively scatter with a compact object, across various mass ratios. We also show that in addition to feedback due to dynamical friction, the dark matter also evolves through a ‘stirring’ effect driven by the time-dependent potential of the binary. We introduce a simple semi-analytical scheme to account for this effect and demonstrate that including stirring tends to slow the rate of dark matter depletion and therefore enhances the impact of dark matter on the dynamics of the binary. Copyright © 2024, The Authors. All rights reserved.

关键词： Gravity waves

来源：评论

学校读者我要写书评

暂无评论

machine learning to detect recent recreational drug use in intensive cardiac care units

引用

Archives of Cardiovascular Diseases 2025年第5期118卷 277-286页

作者： El Bèze, Nathan Hamzi, Kenza Henry, Patrick Trimaille, Antonin El Ouahidi, Amine Zakine, Cyril Nallet, Olivier Delmas, Clément Aboyans, Victor Goralski, Marc Albert, Franck Bonnefoy-Cudraz, Eric Bochaton, Thomas Schurtz, Guillaume Lim, Pascal Lequipar, Antoine Gonçalves, Trecy Gall, Emmanuel Pommier, Thibaut Lemarchand, Léo Meune, Christophe Azzakani, Sonia Bouleti, Claire Amar, Jonas Dillinger, Jean-Guillaume Steg, P. Gabriel Vicaut, Eric Toupin, Solenn Pezel, Théo Inserm MASCOT – UMRS 942 Department of Cardiology University Hospital of Lariboisière Université Paris-Cité AP–HP Paris 75010 France Multimodality Imaging Research for Analysis Core Laboratory: Artificial Intelligence (MIRACL.ai) Department of Data Science Machine Learning and Artificial Intelligence in Health University Hospital of Lariboisière AP–HP Paris 75010 France Department of Cardiovascular Medicine Nouvel Hôpital Civil Strasbourg University Hospital Strasbourg 67000 France Department of Cardiology University Hospital of Brest Brest 29609 France NCT+ Saint-Cyr-sur-Loire 37540 France Department of Cardiology Hôpital Montfermeil Montfermeil 93370 France Department of Cardiology Rangueil University Hospital Toulouse 31000 France Department of Cardiology University Hospital of Limoges Limoges 87000 France Department of Cardiology Centre Hospitalier d'Orléans Orléans 45100 France Department of Cardiology Centre Hospitalier de Chartres Le Coudray 28630 France Intensive Cardiological Care Division Louis-Pradel Hospital Hospices Civils de Lyon Bron 69500 France Department of Cardiology University Hospital of Lille Lille 59000 France Intensive Cardiac Care Unit Henri-Mondor University Hospital Créteil 94000 France Department of Cardiology Dijon University Hospital Dijon 21000 France Department of Cardiology and Vascular Diseases CHU of Rennes Rennes 35000 France Department of Cardiology Hôpital Avicenne AP–HP Bobigny 93000 France Department of Cardiology Clinical Investigation Centre (Inserm 1204) University Hospital of Poitiers Poitiers 86000 France Inserm_U1148/LVTS hôpital Bichat université Paris-Cité AP–HP Paris 75877 France Unité de recherche clinique hôpital Fernand-Widal AP–HP Paris 75010 France

Background: Although recreational drug use is a strong risk factor for acute cardiovascular events, systematic testing is currently not performed in patients admitted to intensive cardiac care units, with a risk of underdetection. To address this issue, machine learning methods could assist in the detection of recreational drug use. Aims: To investigate the accuracy of a machine learning model using clinical, biological and echocardiographic data for detecting recreational drug use in patients admitted to intensive cardiac care units. Methods: From 07 to 22 April 2021, systematic screening for all traditional recreational drugs (cannabis, opioids, cocaine, amphetamines, 3,4-methylenedioxymethamphetamine) was performed by urinary testing in all consecutive patients admitted to intensive cardiac care units in 39 French centres. The primary outcome was recreational drug detection by urinary testing. The framework involved automated variable selection by eXtreme Gradient Boosting (XGBoost) and model building with multiple algorithms, using 31 centres as the derivation cohort and eight other centres as the validation cohort. Results: Among the 1499 patients undergoing urinary testing for drugs (mean age 63 ± 15 years;70% male), 161 (11%) tested positive (cannabis: 9.1%;opioids: 2.1%;cocaine: 1.7%;amphetamines: 0.7%;3,4-methylenedioxymethamphetamine: 0.6%). Of these, only 57% had reported drug use. Using nine variables, the best machine learning model (random forest) showed good performance in the derivation cohort (area under the receiver operating characteristic curve = 0.82) and in the validation cohort (area under the receiver operating characteristic curve = 0.76). Conclusions: In a large intensive cardiac care unit cohort, a comprehensive machine learning model exhibited good performance in detecting recreational drug use, and provided valuable insights into the relationships between clinical variables and drug use through explainable machine learning techniques. ©

关键词： Intensive cardiac care machine learning Recreational drug use

来源：评论

学校读者我要写书评

暂无评论

Automating airborne pollen classification: Identifying and interpreting hard samples for classifiers

引用

Heliyon 2025年第2期11卷 e41656页

作者： Milling, Manuel Rampp, Simon D.N. Triantafyllopoulos, Andreas Plaza, Maria P. Brunner, Jens O. Traidl-Hoffmann, Claudia Schuller, Björn W. Damialis, Athanasios CHI – Chair of Health Informatics MRI Technical University of Munich Munich Germany MCML–Munich Center for Machine Learning Germany EIHW – Chair of Embedded Intelligence for Health Care & Wellbeing University of Augsburg Augsburg Germany Institute of Environmental Medicine and Integrative Health Faculty of Medicine University Clinic of Augsburg & University of Augsburg Augsburg Germany Institute of Environmental Medicine Helmholtz Center Munich German Research Center for Environmental Health Germany Faculty of Business and Economics and Faculty of Medicine University of Augsburg Augsburg Germany Department of Technology Management and Economics Technical University of Denmark Denmark Next Generation Technology Region Zealand Denmark Christine Kühne Center for Allergy Research and Education Davos Switzerland MDSI–Munich Data Science Institute Germany GLAM–the Group on Language Audio & Music Imperial College London London United Kingdom Terrestrial Ecology and Climate Change Department of Ecology School of Biology Faculty of Sciences Aristotle University of Thessaloniki Thessaloniki Greece

Deep-learning-based classification of pollen grains has been a major driver towards automatic monitoring of airborne pollen. Yet, despite an abundance of available datasets, little effort has been spent to investigate which aspects pose the biggest challenges to the (often black-box- resembling) pollen classification approaches. To shed some light on this issue, we conducted a sample-level difficulty analysis based on the likelihood for one of the largest automatically-generated datasets of pollen grains on microscopy images and investigated the reason for which certain airborne samples and specific pollen taxa pose particular problems to deep learning algorithms. It is here concluded that the main challenges lie in A) the (partly) co-occurring of multiple pollen grains in a single image, B) the occlusion of specific markers through the 2D capturing of microscopy images, and C) for some taxa, a general lack of salient, unique features. Our code is publicly available under https://***/millinma/SDPollen © 2025

关键词： Deep learning Pollen recognition Sample difficulty analysis

来源：评论

学校读者我要写书评

暂无评论

Explaining bayesian neural networks

arXiv

引用

arXiv 2021年

作者： Bykov, Kirill Höhne, Marina M.-C. Creosteanu, Adelaida Müller, Klaus-Robert Machine Learning Group Technische Universität Berlin Marchstr. 23 Berlin10587 Germany Department of Artificial Intelligence Korea University Anam-dong Seongbuk-gu Seoul02841 Korea Republic of Max Planck Institute for Informatics Stuhlsatzenhausweg 4 Saarbrücken66123 Germany BIFOLD - Berlin Institute for the Foundations of Learning and Data Technische Universität Berlin Berlin Germany Google Research Brain team Berlin Germany Department of Computer Science TU Kaiserslautern Germany Heidelberg Germany Institute of Pathology Charite – Universitätsmedizin Berlin Berlin Germany Aignostics Berlin Germany RIKEN AIP 1-4-1 Nihonbashi Chuo-ku Tokyo Japan

—To make advanced learning machines such as Deep Neural Networks (DNNs) more transparent in decision making, explainable AI (XAI) aims to provide interpretations of DNNs’ predictions. These interpretations are usually given in the form of heatmaps, each one illustrating relevant patterns regarding the prediction for a given instance. Bayesian approaches such as Bayesian Neural Networks (BNNs) so far have a limited form of transparency (model transparency) already built-in through their prior weight distribution, but notably, they lack explanations of their predictions for given instances. In this work, we bring together these two perspectives of transparency into a holistic explanation framework for explaining BNNs. Within the Bayesian framework, the network weights follow a probability distribution. Hence, the standard (deterministic) prediction strategy of DNNs extends in BNNs to a predictive distribution, and thus the standard explanation extends to an explanation distribution. Exploiting this view, we uncover that BNNs implicitly employ multiple heterogeneous prediction strategies. While some of these are inherited from standard DNNs, others are revealed to us by considering the inherent uncertainty in BNNs. Our quantitative and qualitative experiments on toy/benchmark data and real-world data from pathology show that the proposed approach of explaining BNNs can lead to more effective and insightful explanations. Copyright © 2021, The Authors. All rights reserved.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

Sharpening the dark matter signature in gravitational waveforms. II. Numerical simulations

引用

Physical Review D 2025年第6期111卷 063071-063071页

作者： Bradley J. Kavanagh Theophanes K. Karydas Gianfranco Bertone Pierfrancesco Di Cintio Mario Pasquato Instituto de Física de Cantabria (IFCA UC-CSIC) Avenue de Los Castros s 39005 Santander Spain Gravitation Astroparticle Physics Amsterdam (GRAPPA) Institute for Theoretical Physics Amsterdam and Delta Institute for Theoretical Physics University of Amsterdam Science Park 904 1098 XH Amsterdam The Netherlands Consiglio Nazionale delle Ricerche Istituto dei Sistemi Complessi (CNR-ISC) via Madonna del Piano 17 50022 Sesto Fiorentino (FI) Italy INAF-Osservatorio Astronomico di Arcetri Largo Enrico Fermi 5 50125 Firenze Italy INFN-Sezione di Firenze Via Giovanni Sansone 1 50022 Sesto Fiorentino Italy Département de Physique Université de Montréal 1375 Avenue Thérèse-Lavoie-Roux Montréal Canada Mila—Quebec Artificial Intelligence Institute 6666 Rue Saint-Urbain Montréal Canada Ciela—Montréal Institute for Astrophysical Data Analysis and Machine Learning Montréal Canada Dipartimento di Fisica e Astronomia Università di Padova Vicolo dell’Osservatorio 5 Padova Italy Istituto Nazionale di Fisica Nucleare Padova Via Marzolo 8 Padova Italy

Future gravitational wave observatories can probe dark matter by detecting the dephasing in the waveform of binary black hole mergers induced by dark matter overdensities. Such a detection hinges on the accurate modeling of the dynamical friction, induced by dark matter on the secondary compact object in intermediate and extreme mass ratio inspirals. In this paper, we introduce nbodyimri, a new publicly available code designed for simulating binary systems within cold dark matter spikes. Leveraging higher particle counts and finer time steps, we validate the applicability of the standard dynamical friction formalism and provide an accurate determination of the maximum impact parameter of particles which can effectively scatter with a compact object, across various mass ratios. We also show that in addition to feedback due to dynamical friction, the dark matter also evolves through a stirring effect driven by the time-dependent potential of the binary. We introduce a simple semianalytical scheme to account for this effect and demonstrate that including stirring tends to slow the rate of dark matter depletion and therefore enhances the impact of dark matter on the dynamics of the binary.

关键词：

来源：评论

学校读者我要写书评

暂无评论

A study on the uncertainty of convolutional layers in deep neural networks

arXiv

引用

arXiv 2020年

作者： Shen, Haojing Chen, Sihong Wang, Ran Big Data Institute College of Computer Science and Software Engineering Guangdong Key Lab. of Intelligent Information Processing Shenzhen University ShenzhenGuangdong518060 China College of Mathematics and Statistics Shenzhen University Shenzhen518060 China Shenzhen Key Laboratory of Advanced Machine Learning and Applications Shenzhen University Shenzhen518060 China

This paper shows a Min-Max property existing in the connection weights of the convolutional layers in a neural network structure, i.e., the LeNet. Specifically, the Min-Max property means that, during the back propagation-based training for LeNet, the weights of the convolutional layers will become far away from their centers of intervals, i.e., decreasing to their minimum or increasing to their maximum. From the perspective of uncertainty, we demonstrate that the Min-Max property corresponds to minimizing the fuzziness of the model parameters through a simplified formulation of convolution. It is experimentally confirmed that the model with the Min-Max property has a stronger adversarial robustness, thus this property can be incorporated into the design of loss function. This paper points out a changing tendency of uncertainty in the convolutional layers of LeNet structure, and gives some insights to the interpretability of convolution. © 2020, CC-BY.

关键词： Convolution

来源：评论

学校读者我要写书评

暂无评论

Synchronized sensor insoles for clinical gait analysis in home-monitoring applications

引用

Current Directions in Biomedical Engineering 2018年第1期4卷 433-437页

作者： Roth, Nils Martindale, Christine F. Gaßner, Heiko Kohl, Zacharias Klucken, Jochen Eskofer, Bjoern M. Machine Learning and Data Analytics Lab. Department of Computer Science Friedrich-Alexander-University Erlangen-Nürnberg (FAU) Erlangen Germany Department of Molecular Neurology University Hospital Erlangen Germany

Wearable sensor systems are of increasing interest in clinical gait analysis. However, little information about gait dynamics of patients under free living conditions is available, due to the challenges of integrating such systems unobtrusively into a patient's everyday live. To address this limitation, new, fully integrated low power sensor insoles are proposed, to target applications particularly in home-monitoring scenarios. The insoles combine inertial as well as pressure sensors and feature wireless synchronization to acquire biomechanical data of both feet with a mean timing offset of 15.0 μs. The proposed system was evaluated on 15 patients with mild to severe gait disorders against the GAITRite® system as reference. Gait events based on the insoles' pressure sensors were manually extracted to calculate temporal gait features such as double support time and double support. Compared to the reference system a mean error of 0.06 s ± 0.06 s and 3.89 % ± 2.61 % was achieved, respectively. The proposed insoles proved their ability to acquire synchronized gait parameters and address the requirements for home-monitoring scenarios, pushing the boundaries of clinical gait analysis. © 2018 Nils Roth et al.

关键词： Synchronization

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：