检索结果-内蒙古大学图书馆

Speeding up Heterogeneous Federated learning with Sequentially Trained Superclients 26

Speeding up Heterogeneous Federated Learning with Sequential...

26th international Conference on pattern recognition / 8th international workshop on Image mining - Theory and Applications (IMTA)

作者： Zaccone, Riccardo Rizzardi, Andrea Caldarola, Debora Ciccone, Marco Caputo, Barbara Politecnico Torino Turin Italy

ISBN: (数字)9781665490627

ISBN: (纸本)9781665490627

Federated learning (FL) allows training machine learning models in privacy-constrained scenarios by enabling the cooperation of edge devices without requiring local data sharing. This approach raises several challenges due to the different statistical distribution of the local datasets and the clients' computational heterogeneity. In particular, the presence of highly non-i.i.d. data severely impairs both the performance of the trained neural network and its convergence rate, increasing the number of communication rounds requested to reach a performance comparable to that of the centralized scenario. As a solution, we propose FedSeq, a novel framework leveraging the sequential training of subgroups of heterogeneous clients, Le. superclients, to emulate the centralized paradigm in a privacy-compliant way. Given a fixed budget of communication rounds, we show that FedSeq outperforms or match several state-of-the-art federated algorithms in terms of final performance and speed of convergence. Finally, our method can be easily integrated with other approaches available in the literature. Empirical results show that combining existing algorithms with FedSeq further improves its final performance and convergence speed. We test our method on CIFAR-10 and CIFAR-100 and prove its effectiveness in both i.i.d. and non-i.i.d. scenarios.(1)

关键词： Training Performance evaluation Federated learning Computational modeling Neural networks statistical distributions data models

来源：评论

学校读者我要写书评

暂无评论

Rainfall Prediction Using machine learning 1

Rainfall Prediction Using Machine Learning

引用

1st international Conference on Artificial Intelligence Trends and pattern recognition, ICAITPR 2022

作者： Gupta, Akash Mall, Hitesh Kumar Janarthanan, S. Galgotias University School of Computing Science and Engineering UP Greater Noida India

ISBN: (数字)9781665425216

ISBN: (纸本)9781665425216

Rainfall forecasting is a single of difficult and unpredictable undertakings that has a major influence on human society. Predictions that are accurate and timely can help to avert human and financial loss. This study contains a series of experiments that include the utilisation of basic machine learning techniques to build Weather forecasting models that estimate whether it will rain in major cities tomorrow based on the day's meteorological data. This comparative research looks at three aspects of modelling inputs, modelling methodologies, and preprocessing procedures. The findings demonstrate how different machine learning systems perform on a range of assessment parameters, as well as their capacity to forecast rainfall using weather data analysis. Agriculture is crucial to India's survival. The importance of rainfall in agriculture cannot be overstated. Rainfall forecasting has been a key problem in recent years. Individuals can be more aware of the weather and make more educated judgments by predicting rainfall. © 2022 IEEE.

关键词： Rain

来源：评论

学校读者我要写书评

暂无评论

Optimal Feature Subset Selection for pattern Classification and recognition Using Spectral Entropy as a Complexity Measure from Alcohol and Sleep EEG data 1

Optimal Feature Subset Selection for Pattern Classification ...

引用

1st IEEE international Conference on Innovations in High Speed Communication and Signal Processing, IHCSP 2023

作者： Singh, Takveer Chitkara University Institute of Engineering and Technology Chitkara University Centre for Interdisciplinary Research in Business and Technology Punjab India

ISBN: (纸本)9798350345957

The quantity of unprocessed data in multi-channel EEG analysis is enormous. Most datasets have some level of noise or redundancy, which could lower the accuracy of any relevant features that are uncovered. This result in a rise in computational complexity and a corresponding rise in the importance of concerns related to memory consumption. Consequently, feature selection has become the matter of significant research in the arenas of machine learning and statistics since it is crucial for efficient analysis of data while needing less memory and time. Based on the findings of this study, selecting features with high class separability from multichannel EEG recordings is an efficient method for performing supervised learning tasks. With this study, we intend to test the following theory: The best feature subset will have features that are substantially linked with the class labels but not connected with each other. Using the same alcoholic EEG database as these other investigations, we created and compared three feature selection algorithms. The purpose here is to use class knowledge in a feature selection technique to enhance the accuracy with which classes may be detected. © 2023 IEEE.

关键词： Feature Selection

来源：评论

学校读者我要写书评

暂无评论

Efficient Large Scale Medical Image dataset Preparation for machine learning Applications 1

引用

1st MICCAI international workshop on data Engineering in Medical Imaging (DEMI)

作者： Denner, stefan Scherer, Jonas Kades, Klaus Bounias, Dimitrios Schader, Philipp Kausch, Lisa Bujotzek, Markus Bucher, Andreas Michael Penzkofer, Tobias Maier-Hein, Klaus German Canc Res Ctr Div Med Image Comp Heidelberg Germany Heidelberg Univ Med Fac Heidelberg Heidelberg Germany Heidelberg Univ Hosp Dept Radiat Oncol Pattern Anal & Learning Grp Heidelberg Germany Natl Ctr Tumor Dis NCT Heidelberg Germany Goethe Univ Frankfurt Univ Hosp Inst Diagnost & Intervent Radiol Frankfurt Germany Charite Univ Med Berlin Dept Radiol Berlin Germany

ISBN: (数字)9783031449925

ISBN: (纸本)9783031449918;9783031449925

In the rapidly evolving field of medical imaging, machine learning algorithms have become indispensable for enhancing diagnostic accuracy. However, the effectiveness of these algorithms is contingent upon the availability and organization of high-quality medical imaging datasets. Traditional Digital Imaging and Communications in Medicine (DICOM) data management systems are inadequate for handling the scale and complexity of data required to be facilitated in machine learning algorithms. This paper introduces an innovative data curation tool, developed as part of the Kaapana (https://***/kaapana/ kaapana) open-source toolkit, aimed at streamlining the organization, management, and processing of large-scale medical imaging datasets. The tool is specifically tailored to meet the needs of radiologists and machine learning researchers. It incorporates advanced search, auto-annotation and efficient tagging functionalities for improved data curation. Additionally, the tool facilitates quality control and review, enabling researchers to validate image and segmentation quality in large datasets. It also plays a critical role in uncovering potential biases in datasets by aggregating and visualizing metadata, which is essential for developing robust machine learning models. Furthermore, Kaapana is integrated within the Radiological Cooperative Network (RACOON), a pioneering initiative aimed at creating a comprehensive national infrastructure for the aggregation, transmission, and consolidation of radiological data across all university clinics throughout Germany. A supplementary video showcasing the tool's functionalities can be accessed at https://***/MICCAI- DEMI2023.

关键词： Medical Imaging data Curation machine learning Kaapana dataset Preperation Quality Control Bias Detection

来源：评论

学校读者我要写书评

暂无评论

international workshop on Continual Semi-Supervised learning: Introduction, Benchmarks and Baselines 1

引用

1st international workshop on Continual Semi-Supervised learning (CSSLat IJCAI)

作者： Shahbaz, Ajmal Khan, Salman Hossain, Mohammad Asiful Lomonaco, Vincenzo Cannons, Kevin Xu, Zhan Cuzzolin, Fabio Oxford Brookes Univ Oxford England Huawei Technol Vancouver BC Canada Univ Pisa Pisa Italy

ISBN: (数字)9783031175879

ISBN: (纸本)9783031175879;9783031175862

The aim of this paper is to formalise a new continual semi-supervised learning (CSSL) paradigm, proposed to the attention of the machine learning community via the IJCAI 2021 international workshop on Continual Semi-Supervised learning (CSSL@IJCAI) (https://***/view/sscl-workshop-ijcai-2021/), with the aim of raising the field's awareness about this problem and mobilising its effort in this direction. After a formal definition of continual semi-supervised learning and the appropriate training and testing protocols, the paper introduces two new benchmarks specifically designed to assess CSSL on two important computer vision tasks: activity recognition and crowd counting. We describe the Continual Activity recognition (CAR) and Continual Crowd Counting (CCC) challenges built upon those benchmarks, the baseline models proposed for the challenges, and describe a simple CSSL baseline which consists in applying batch self-training in temporal sessions, for a limited number of rounds. The results show that learning from unlabelled data streams is extremely challenging, and stimulate the search for methods that can encode the dynamics of the data stream.

关键词： Continual learning Semi-supervised learning Artificial intelligence

来源：评论

学校读者我要写书评

暂无评论

Knowledge Graph data Enrichment Based on a Software Library for Text Mapping to the Sustainable Development Goals 2

Knowledge Graph Data Enrichment Based on a Software Library ...

引用

Joint of the 2nd international workshop on Knowledge Graph Generation From Text and the 1st international BiKE Challenge, TEXT2KG 2023 and BiKE 2023

作者： Mandilara, Ioanna Fotopoulou, Eleni Androna, Christina Maria Zafeiropoulos, Anastasios Papavassiliou, Symeon Institute of Communication and Computer Systems National Technical University of Athens Athens Greece

Over the last few years, there has been a significant increase in the release of massive amounts of data related to the Sustainable Development Goals (SDGs). This has been driven by the recognition that data is critical for monitoring progress towards the SDGs, identifying areas that require attention, and informing policy decisions. Given that a significant percentage of the provided information is made available in documents, the availability of software libraries that can enable scientists to easily extract information regarding the importance given to the SDGs in the documents’ text is considered crucial. Towards this direction, we have developed an open-source Python software library that provides the mapping between text and the SDGs, taking advantage of novel machine learning techniques. The provided software library is modular and permits the dynamic selection of trained machine learning models for analysis purposes. The outcomes from the usage of the software library are fed for data population of a knowledge graph that is targeted to the tracking of information around the SDGs. The overall approach is open and can be easily adopted by scientists and policy makers to support participatory modeling processes, as well as participatory decision making and action planning for the development of solutions for climate-resilient regions. © 2023 CEUR-WS. All rights reserved.

关键词： Sustainable development

来源：评论

学校读者我要写书评

暂无评论

Cross-Modality Image Quality Prediction for Time-Resolved CT from Breathing Signals

Cross-Modality Image Quality Prediction for Time-Resolved C...

引用

workshop on Longitudinal Disease Tracking and Modeling with Medical Images and data, LDTM 2024, 5th international workshop on Multiscale Multimodal Medical Imaging, MMMI 2024, 1st workshop on machine learning for Multimodal/-sensor Healthcare data, ML4MHD2024 and workshop on Multimodal learning and Fusion Across Scales for Clinical Decision Support, ML-CDS 2024 held in conjunction with the 27th international Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2024

作者： Schwarz, Annette Dickmann, Jannis Hofmann, Christian Szkitsak, Juliane Bert, Christoph Maier, Andreas Arias-Vergara, Tomás Pattern Recognition Lab Friedrich-Alexander-Universität Erlangen-Nürnberg Erlangen Germany Siemens Healthineers AG Forchheim Germany Department of Radiation Oncology Universitätsklinikum Erlangen Friedrich-Alexander-Universität Erlangen-Nürnberg Erlangen Germany Erlangen Germany

ISBN: (纸本)9783031845246

Four-dimensional computed tomography (4DCT) is a time-resolved, multi-modal imaging method that captures respiratory signals synchronised with the CT scan in order to track the movement of the lung. It is routinely used in radiation therapy treatment planning for lung or liver cancer patients. However, image artifacts during 4DCT scans caused by irregular patient breathing negatively impact treatment outcomes. This work proposes a method to automatically detect patients at high risk for severe image artifacts even before the scan is conducted based on an pre-scan analysis of their breathing. This can help to take proactive measures to improve image quality, such as changing the scan mode or providing in-depth patient coaching. A deep neural network is trained to predict the image quality score of 28 lung CT phantom scans, each rated by ten clinical experts. Different pretrained networks are investigated for feature generation and combined with two linear output heads to predict the average expert image quality score of unseen scans. We were able to predict the quality of a 4DCT with a mean absolute error of 8% using only the one-dimensional breathing signal as input. This accuracy is comparable to the rating consistency of our clinical experts, which were rating the images directly. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Radiotherapy

来源：评论

学校读者我要写书评

暂无评论

Sensor Fusion and Multimodal learning for Robotic Grasp Verification Using Neural Networks 26

Sensor Fusion and Multimodal Learning for Robotic Grasp Veri...

引用

26th international Conference on pattern recognition / 8th international workshop on Image mining - Theory and Applications (IMTA)

作者： Gohil, Priteshkumar Thoduka, Santosh Ploeger, Paul G. Bonn Rhein Sieg Univ Appl Sci Dept Comp Sci Autonomous Syst St Augustin Germany

ISBN: (数字)9781665490627

ISBN: (纸本)9781665490627

Different sensors on a robot help in understanding different aspects of the environment they are working in;however, each sensor modality is often processed individually and information from other sensors is not utilized jointly. One of the reasons is different sampling rates and different dimensions of input modalities. In this paper, we use multimodal data fusion techniques such as early, late and intermediate fusion for grasp failure identification using four different 3D convolution-based multimodal neural networks (3D-MNN). Our results on a visualtactile dataset shows that the performance of the classification task is improved while using multimodal data. In addition, a neural network trained with 30:22 train-test split of multimodal data achieved accuracy comparable to a network trained with 78:22 train-test split of unimodal data(1).

关键词： Visualization Three-dimensional displays Fuses Neural networks Sensor fusion Robot sensing systems Minimization

来源：评论

学校读者我要写书评

暂无评论

SHM data anomaly classification using machine learning strategies: A comparative study

引用

SMART stRUCTURES AND SYstEMS 2022年第1期29卷 77-91页

作者： Chou, Jau-Yu Fu, Yuguang Huang, Shieh-Kung Chang, Chia-Ming Natl Taiwan Univ Dept Civil Engn 1 Sec 4Roosevelt Rd Taipei 10617 Taiwan Nanyang Technol Univ Sch Civil & Environm Engn 50 Nanyang Ave Singapore 639798 Singapore Natl Chung Hsing Univ Dept Civil Engn 145 Xingda Rd Taichung 40227 Taiwan

Various monitoring systems have been implemented in civil infrastructure to ensure structural safety and integrity. In long-term monitoring, these systems generate a large amount of data, where anomalies are not unusual and can pose unique challenges for structural health monitoring applications, such as system identification and damage detection. Therefore, developing efficient techniques is quite essential to recognize the anomalies in monitoring data. In this study, several machine learning techniques are explored and implemented to detect and classify various types of data anomalies. A field dataset, which consists of one month long acceleration data obtained from a long-span cable-stayed bridge in China, is employed to examine the machine learning techniques for automated data anomaly detection. These techniques include the statistic-based pattern recognition network, spectrogram-based convolutional neural network, image-based time history convolutional neural network, image-based time-frequency hybrid convolution neural network (GoogLeNet), and proposed ensemble neural network model. The ensemble model deliberately combines different machine learning models to enhance anomaly classification performance. The results show that all these techniques can successfully detect and classify six types of data anomalies (i.e., missing, minor, outlier, square, trend, drift). Moreover, both image-based time history convolutional neural network and GoogLeNet are further investigated for the capability of autonomous online anomaly classification and found to effectively classify anomalies with decent performance. As seen in comparison with accuracy, the proposed ensemble neural network model outperforms the other three machine learning techniques. This study also evaluates the proposed ensemble neural network model to a blind test dataset. As found in the results, this ensemble model is effective for data anomaly detection and applicable for the signal characteristics changi

关键词： data anomaly classification ensemble neural network GoogLeNet machine learning

来源：评论

学校读者我要写书评

暂无评论

Explanation-based Graph Neural Networks for Graph Classification 26

Explanation-based Graph Neural Networks for Graph Classifica...

引用

26th international Conference on pattern recognition / 8th international workshop on Image mining - Theory and Applications (IMTA)

作者： Seo, Sangwoo Jung, Seungjun Kim, Changick Korea Adv Inst Sci & Technol KAIST Sch Elect Engn 291 Daehak Ro Daejeon 34141 South Korea

ISBN: (数字)9781665490627

ISBN: (纸本)9781665490627

Graph Neural Network models can be used to quickly analyze interactions between multiple data expressed in a graph structure, with high accuracy. Previous studies accurately extract subgraphs which have a significant influence on the whole graph, providing accurate explanations for predictions of GNN. We noted that explanation components could help improve classification performance as unique representations of each class. Therefore, we suggest the GNN performance can be further improved by using explanation components. In this paper, we propose an Explanation-Based Graph Neural Networks (EBGNN) that utilizes contrastive learning at the instance level, by applying explanation components. In EBGNN, the explanation components ensure similarity for instances within the same class, and promote separability for instances in different classes. Finally, we conducted an evaluation on five benchmark datasets (MUTAG, IMDB-BINARY, PROTEINS, NCI1, and DD). Our experiment showed a significant increase in graph classification performance compared to state-of-the-art methods.

关键词： Proteins Computer vision Analytical models machine learning Benchmark testing Graph neural networks data models

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：