检索结果-内蒙古大学图书馆

IEEE Conference on Decision and Control

作者： Hao Yuan Sulaiman A. Alghunaim Kun Yuan Center for Machine Learning Research Peking University Beijing P. R. China Dept. Electrical Engr. Kuwait University Safat Kuwait AI for Science Institute Beijing P. R. China National Engineering Labratory for Big Data Analytics and Applications Beijing P. R. China

Decentralized stochastic optimization has become a crucial tool for addressing large-scale machine learning and control problems. In decentralized algorithms, all computing nodes are connected through a network topology, and each node communicates only with its direct neighbors. Decentralized algorithms can significantly reduce communication overhead by eliminating the need for global communication. However, existing research on the linear speedup analysis of decentralized stochastic algorithms is limited to the condition of network-dependent learning rates, which rarely holds in practice since the network connectivity is typically unknown to each node. As a result, it remains an open question whether a linear speedup bound can be achieved using network-independent learning rates. This paper provides an affirmative answer. By utilizing a new analysis framework, we prove that D-SGD and Exact-Diffusion, two representative decentralized stochastic algorithms, can achieve linear speedup with network-independent learning rates. Simulations are provided to validate our theories.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Don’t Get Me Wrong: How to Apply Deep Visual Interpretations to Time Series

arXiv

引用

arXiv 2022年

作者： Löffler, Christoffer Lai, Wei-Cheng Eskofier, Björn M. Zanca, Dario Schmidt, Lukas Mutschler, Christopher Machine Learning and Data Analytics Lab Friedrich-Alexander University Erlangen-Nuremberg Carl-Thiersch-Straße 2b Erlangen91052 Germany Fraunhofer IIS Fraunhofer Institute for Integrated Circuits IIS Positioning and Networks Division Erlangen Germany

The correct interpretation and understanding of deep learning models are essential in many applications. Explanatory visual interpretation approaches for image, and natural language processing allow domain experts to validate and understand almost any deep learning model. However, they fall short when generalizing to arbitrary time series, which is inherently less intuitive and more diverse. Whether a visualization explains valid reasoning or captures the actual features is difficult to judge. Hence, instead of blind trust, we need an objective evaluation to obtain trustworthy quality metrics. We propose a framework of six orthogonal metrics for gradient-, propagation- or perturbation-based post-hoc visual interpretation methods for time series classification and segmentation tasks. An experimental study includes popular neural network architectures for time series and nine visual interpretation methods. We evaluate the visual interpretation methods with diverse datasets from the UCR repository and a complex, real-world dataset and study the influence of standard regularization techniques during training. We show that none of the methods consistently outperforms others on all metrics, while some are sometimes ahead. Our insights and recommendations allow experts to choose suitable visualization techniques for the model and task. © 2022, CC BY.

关键词： Time series

来源：评论

学校读者我要写书评

暂无评论

The EmpkinS-EKSpression Reappraisal Training Augmented With Kinesthesia in Depression: One-Armed Feasibility Study

引用

JMIR Formative Research 2025年 9卷 e65357页

作者： Keinert, Marie Schindler-Gmelch, Lena Rupp, Lydia Helene Sadeghi, Misha Richer, Robert Capito, Klara Eskofier, Bjoern M. Berking, Matthias Department of Clinical Psychology and Psychotherapy Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU) Erlangen Germany Machine Learning and Data Analytics Lab Department of Artificial Intelligence in Biomedical Engineering (AIBE) Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU) Erlangen Germany Translational Digital Health Group Institute of AI for Health Helmholtz Zentrum Muenchen Neuherberg Germany

Background: Harboring dysfunctional depressogenic cognitions contributes to the development and maintenance of depression. A central goal of cognitive behavioral therapy (CBT) for depression is to invalidate such cognitions via cognitive reappraisal (CR). However, relatively low remission rates and high dropout rates in CBT demonstrate the need for further improvement. Potentially, the effects of CBT could be enhanced by addressing not only dysfunctional depressogenic cognitions but also body states associated with depression. This may be done, for example, by systematically pairing the invalidation of depressogenic cognitions with the performance of antidepressive kinesthesia. Objective: This study aimed to examine the feasibility and clinical potential of a smartphone-based cognitive restructuring task that required users to deliberately perform antidepressive kinesthesia in conjunction with the rejection of depressogenic statements and the affirmation of antidepressive statements. This feasibility study was conducted as a precursor to a large-scale randomized controlled trial. Methods: In total, 10 healthy participants engaged in a single 90-120-minute session of smartphone-based CR training. During the training, they completed 2 phases in which they were required to reject 20 depressogenic and affirm 20 antidepressive statements, respectively. Diagnostic assessments were conducted 1 week (T1) before and directly prior (T2) to the training, and again directly posttraining (T3) and at a 2-week follow-up posttraining (T4). Feasibility outcomes assessed at T3 included intervention safety recorded by study therapists, compliance, technical feasibility, usability assessed using the Short Version of the User Experience Questionnaire (UEQ-S), and acceptability assessed using the UEQ-S and self-developed items. Preliminary clinical potential was evaluated via single-item ratings of current depressed and positive mood assessed continuously during the training. Feasibility

关键词： cognitive reappraisal depression facial expression kinesthesia mobile phone smartphone-based intervention

来源：评论

学校读者我要写书评

暂无评论

Language-Agnostic Bias Detection in Language Models with Bias Probing

arXiv

引用

arXiv 2023年

作者： Koksal, Abdullatif Yalcin, Omer Faruk Akbiyik, Ahmet Kilavuz, M. Tahir Korhonen, Anna Schutze, Hinrich Center for Information and Language Processing LMU Munich Germany Munich Center for Machine Learning Germany Data Analytics and Computational Social Science University of Massachusetts Amherst United States Harvard Kennedy School United States Middle East Initiative Harvard Kennedy School United States Marmara University Turkey Language Technology Lab University of Cambridge United Kingdom

Pretrained language models (PLMs) are key components in NLP, but they contain strong social biases. Quantifying these biases is challenging because current methods focusing on fill-the-mask objectives are sensitive to slight changes in input. To address this, we propose a bias probing technique called LABDet, for evaluating social bias in PLMs with a robust and language-agnostic method. For nationality as a case study, we show that LABDet "surfaces" nationality bias by training a classifier on top of a frozen PLM on non-nationality sentiment detection. We find consistent patterns of nationality bias across monolingual PLMs in six languages that align with historical and political context. We also show for English BERT that bias surfaced by LABDet correlates well with bias in the pretraining data;thus, our work is one of the few studies that directly links pretraining data to PLM behavior. Finally, we verify LABDet's reliability and applicability to different templates and languages through an extensive set of robustness checks. We publicly share our code and dataset in https://***/akoksal/LABDet. Copyright © 2023, The Authors. All rights reserved.

关键词： machine learning

来源：评论

学校读者我要写书评

暂无评论

T-Cell Receptor Optimization with Reinforcement learning and Mutation Polices for Precision Immunotherapy 27th

T-Cell Receptor Optimization with Reinforcement Learning an...

引用

27th International Conference on Research in Computational Molecular Biology, RECOMB 2023

作者： Chen, Ziqi Min, Martin Renqiang Guo, Hongyu Cheng, Chao Clancy, Trevor Ning, Xia Computer Science and Engineering The Ohio State University ColumbusOH43210 United States Machine Learning Department NEC Labs PrincetonNJ08540 United States Digital Technologies Research Centre National Research Council Canada Ontario Canada Department of Medicine Baylor College of Medicine HoustonTX77030 United States NEC Oncolmmunity AS Oslo Cancer Cluster Innovation Park Ullernchausséen 64 Oslo0379 Norway Biomedical Informatics The Ohio State University ColumbusOH43210 United States Translational Data Analytics Institute The Ohio State University ColumbusOH43210 United States

ISBN: (纸本)9783031291180

T cells monitor the health status of cells by identifying foreign peptides displayed on their surface. T-cell receptors (TCRs), which are protein complexes found on the surface of T cells, are able to bind to these peptides. This process is known as TCR recognition and constitutes a key step for immune response. Optimizing TCR sequences for TCR recognition represents a fundamental step towards the development of personalized treatments to trigger immune responses killing cancerous or virus-infected cells. In this paper, we formulated the search for these optimized TCRs as a reinforcement learning (Formula Presented) problem, and presented a framework (Formula Presented) with a mutation policy using proximal policy optimization. (Formula Presented) mutates TCRs into effective ones that can recognize given peptides. (Formula Presented) leverages a reward function that combines the likelihoods of mutated sequences being valid TCRs measured by a new scoring function based on deep autoencoders, with the probabilities of mutated sequences recognizing peptides from a peptide-TCR interaction predictor. We compared (Formula Presented) with multiple baseline methods and demonstrated that (Formula Presented) significantly outperforms all the baseline methods to generate positive binding and valid TCRs. These results demonstrate the potential of(Formula Presented) for both precision immunotherapy and peptide-recognizing TCR motif discovery. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Automated Pipeline for Regional Epicardial Adipose Tissue Distribution Analysis in the Left Atrium 15th

Automated Pipeline for Regional Epicardial Adipose Tissue Di...

引用

15th International Workshop on Statistical Atlases and Computational Models of the Heart, STACOM 2024, Held in Conjunction with MICCAI 2024

作者： Lluch, Èric Castañeda, Eduardo Weiss, Maximilian E. R. Vizitiu, Anamaria Jacob, Athira Jami, Jitin Audigier, Chloé Meister, Felix Mihalef, Viorel Passerini, Tiziano Siemens Healthineers Digital Technologies and Innovation Erlangen Germany Pattern Recognition Lab Department of Computer Science Friedrich-Alexander University Erlangen-Nürnberg Erlangen Germany Siemens SRL Brasov Romania Siemens Healthineers Digital Technology and Innovation Princeton United States Machine-Learning and Data Analytics Friedrich-Alexander-Universität Erlangen-Nürnberg Erlangen Germany Advanced Clinical Imaging Technology Siemens Healthineers AG Lausanne Switzerland

ISBN: (纸本)9783031877551

Atrial fibrillation (AF) is the most common cardiac arrhythmia, affecting approximately 3% of the global population and rising to 11% in individuals over 80 years old. The distribution of epicardial adipose tissue (EAT) around the atria plays a significant role in the pathophysiology of AF, yet its regional analysis remains underexplored. This paper introduces an automated pipeline capable of segmenting 14 atrial regions from patient CT angiograms, a notable advancement over previous attempts to automate the creation of synthetic anatomical or physics-based twins. Our fully automatic pipeline requires no human supervision and provides the regional EAT quantification successfully on 91.9% of the analyzed patients for which the LA and its structures were automatically segmented (1797 out of 1954 patients), including the automatic segmentation of the aorta and pericardium to assess EAT volume around the left atrium. By applying principal component analysis of the regional EAT to analyze a separate dataset of 84 individuals who were clinically labeled, we identified a distinct group that contained all the patients with AF. This finding suggests that there is a significant link between the pattern of fat distribution and AF. This novel approach has the potential to enhance our understanding of AF and improve patient outcomes by enabling large-scale, regional fat analysis in an automated manner on all patients with accurate segmentation of left atrial structures. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Principal component analysis

来源：评论

学校读者我要写书评

暂无评论

Exploration of Process Mining Opportunities In Educational Software Engineering - The GitLab Analyser 13

Exploration of Process Mining Opportunities In Educational S...

引用

13th International Conference on Educational data Mining, EDM 2020

作者： Dumbach, Philipp Aly, Alexander Zrenner, Markus Eskofier, Bjoern M. Machine Learning and Data Analytics Lab Department of Computer Science Friedrich-Alexander-Universität Erlangen-Nürnberg Carl-Thiersch-Str. 2b Erlangen91052 Germany

ISBN: (纸本)9781733673617

The increasing complexity in software development leads to the necessity for a detailed data analysis. Literature illustrates a stronger research focus on Educational Process Mining (EPM) being applied to the fields of e-learning and professional training. In this work, the opportunities of Process Mining (PM) are further examined by the evaluation of software engineering (SE) courses. The methodology follows the five stages of the L* life cycle model for PM projects using data from software repositories. The event log data was analyzed with the PM tool Disco to examine the students’ work following an agile development process. The new tool GitLab Analyser supports supervisors to visualize educational processes and still extracts event logs for the further analysis and application of PM techniques. © 2020 Proceedings of the 13th International Conference on Educational data Mining, EDM 2020. All rights reserved.

关键词： data mining

来源：评论

学校读者我要写书评

暂无评论

Spectral, Probabilistic, and Deep Metric learning: Tutorial and Survey

arXiv

引用

arXiv 2022年

作者： Ghojogh, Benyamin Ghodsi, Ali Karray, Fakhri Crowley, Mark Department of Electrical and Computer Engineering Machine Learning Laboratory University of Waterloo WaterlooON Canada Department of Statistics and Actuarial Science David R. Cheriton School of Computer Science Data Analytics Laboratory University of Waterloo WaterlooON Canada Department of Electrical and Computer Engineering Centre for Pattern Analysis and Machine Intelligence University of Waterloo WaterlooON Canada

This is a tutorial and survey paper on metric learning. Algorithms are divided into spectral, probabilistic, and deep metric learning. We first start with the definition of distance metric, Mahalanobis distance, and generalized Mahalanobis distance. In spectral methods, we start with methods using scatters of data, including the first spectral metric learning, relevant methods to Fisher discriminant analysis, Relevant Component Analysis (RCA), Discriminant Component Analysis (DCA), and the Fisher-HSIC method. Then, large-margin metric learning, imbalanced metric learning, locally linear metric adaptation, and adversarial metric learning are covered. We also explain several kernel spectral methods for metric learning in the feature space. We also introduce geometric metric learning methods on the Riemannian manifolds. In probabilistic methods, we start with collapsing classes in both input and feature spaces and then explain the neighborhood component analysis methods, Bayesian metric learning, information theoretic methods, and empirical risk minimization in metric learning. In deep learning methods, we first introduce reconstruction autoencoders and supervised loss functions for metric learning. Then, Siamese networks and its various loss functions, triplet mining, and triplet sampling are explained. Deep discriminant analysis methods, based on Fisher discriminant analysis, are also reviewed. Finally, we introduce multi-modal deep metric learning, geometric metric learning by neural networks, and few-shot metric learning. Copyright © 2022, The Authors. All rights reserved.

关键词： Surveys

来源：评论

学校读者我要写书评

暂无评论

Radar-based Recognition of Activities of Daily Living in the Palliative Care Context Using Deep learning

Radar-based Recognition of Activities of Daily Living in the...

引用

IEEE EMBS International Conference on Information Technology Applications in Biomedicine (ITAB)

作者： Johanna Braeunig Desar Mejdani Daniel Krauss Stefan Griesshammer Robert Richer Christian Schuessler Julia Yip Tobias Steigleder Christoph Ostgathe Bjoern M. Eskofier Martin Vossiek Department Electrical Engineering Institute of Microwaves and Photonics Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU) Erlangen Germany Department Artificial Intelligence in Biomedical Engineering (AIBE) Machine Learning and Data Analytics Lab (MaD Lab) Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU) Erlangen Germany Department of Palliative Medicine Universitätsklinikum Erlangen Erlangen Germany

The accurate detection and quantification of activities of daily life (ADL) are crucial for assessing the health status of palliative patients to allow an optimized treatment in the last phase of life. Current evaluation methods heavily rely on subjective self-reports or external observations by clinical staff, lacking objectivity. To address this limitation, we propose a radar-based approach for recognizing ADLs in a palliative care context. In our proof of concept study, we recorded five different ADLs relevant to palliative care, all occurring within a hospital bed, from N=14 healthy participants (57% women, aged 28.6 ± 5.3years). All movements were recorded using two frequency-modulated continuous wave radar systems measuring velocity, range, and angle. A convolutional neural network combined with long short-term memory achieved a classification accuracy of 99.8 ± 0.4% across five cross-validation folds. Furthermore, we compare our initial approach, which takes into account all dimensions of the available radar data, to a simplified version, where only velocity information over time is fed into the network. While these results demonstrate the high potential of radar-based sensing to automatically detect and quantify activities in a palliative care context, future work is still necessary to assess the applicability to real-world hospital scenarios.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Impact of Distributed Training on Mask R-CNN Model Performance for Image Segmentation 1

Impact of Distributed Training on Mask R-CNN Model Performan...

引用

1st International Conference on Emerging Trends in Information Technology and Engineering, ic-ETITE 2020

作者： Ranjit, Mercy Prasanna Ganapathy, Gopinath Manuel, Ranjit Frederick Microsoft Corporation India Ltd AI Machine Learning Bangalore India Bharathidasan University School of Computer Science and Engineering Trichy India ITC Infotech India Ltd Data Analytics AI Bangalore India

ISBN: (纸本)9781728141428

The paper compares and discusses the impact of distributed deep learning training using the Horovod framework on the performance metrics of image segmentation model trained using the Mask R-CNN (Region-based Convolutional Neural Network) architecture in comparison with the baseline metrics obtained from running the training on a single node. Image segmentation is a complex problem in comparison with image classification and the fluctuations introduced due to the distributed gradient computations while running the training distributed is less pronounced on simpler problems like image classification while it is more pronounced for image segmentation where we measure the performance at pixel level. It is important to know how the latest algorithms and distribution frameworks perform when training the model on distributed compute with parallel gradient calculations, adjusted learning rates and averaged neural network weight updates. The paper also aimed to understand if the effective batch size per training iteration while performing the training distributed did not negatively impact the ability of the model to generalize when increasing the batch size as indicated in the previous studies. The distributed training with Horovod was performed using the Azure machine learning service on scalable cloud compute cluster. The aim of the study was not comparison in-terms of the execution time which has been well established on previous studies but on understanding the impact on the quality of the predictions measured with standard validation metrics for image segmentation using the ISIC (International Skin Imaging Collaboration) 2018 Skin Lesion image dataset. © 2020 IEEE.

关键词： Image segmentation

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：