检索结果-内蒙古大学图书馆

arXiv 2022年

作者： Dablain, Damien A. Bellinger, Colin Krawczyk, Bartosz Aha, David W. Chawla, Nitesh V. Lucy Family Institute for Data & Society Dept. Computer Science and Engineering University of Notre Dame Notre DameIN46556 United States National Research Council of Canada OttawaK1A 0R6 Canada Dept. Computer Science Virginia Commonwealth University RichmondVA23824 United States Navy Center for Applied Research in AI Naval Research Laboratory WashingtonDC20375 United States

Deep learning models are being increasingly applied to imbalanced data in high stakes fields such as medicine, autonomous driving, and intelligence analysis. Imbalanced data compounds the black-box nature of deep networks because the relationships between classes may be highly skewed and unclear. This can reduce trust by model users and hamper the progress of developers of imbalanced learning algorithms. Existing methods that investigate imbalanced data complexity are geared toward binary classification, shallow learning models and low dimensional data. In addition, current eXplainable Artificial Intelligence (XAI) techniques mainly focus on converting opaque deep learning models into simpler models (e.g., decision trees) or mapping predictions for specific instances to inputs, instead of examining global data properties and complexities. Therefore, there is a need for a framework that is tailored to modern deep networks, that incorporates large, high dimensional, multi-class datasets, and uncovers data complexities commonly found in imbalanced data (e.g., class overlap, sub-concepts, and outlier instances). We propose a set of techniques that can be used by both deep learning model users to identify, visualize and understand class prototypes, sub-concepts and outlier instances;and by imbalanced learning algorithm developers to detect features and class exemplars that are key to model performance. Our framework also identifies instances that reside on the border of class decision boundaries, which can carry highly discriminative information. Unlike many existing XAI techniques which map model decisions to gray-scale pixel locations, we use saliency through back-propagation to identify and aggregate image color bands across entire classes. This facilitates the high-level analysis of foreground and background colors used by models to discern classes (e.g., blue sky for airplanes). Our framework is publicly available at https://***/dd1github/XAI_for_Imbalanced_

关键词： Decision trees

来源：评论

学校读者我要写书评

暂无评论

Cost optimization in CNN for big data of share price

引用

AIP Conference Proceedings 2023年第1期2587卷

作者： R. Saravanakumar Gokul Talele Ali Hazim Alyamoor Kaushal Kishor Bhatt Prashant Vats Malik Bader Alazzam Dept. of Computer Science and Engineering Dayananda Sagar Academy of Technology and Management Bangalore India Dept. of Data Science IIIT Banglore India Dept. of Accounting Economic and Administration College University of Mosul Ninawa Iraq Dept. of Computer Science and Engineering Birla Institute of Applied Science Bhimtal Nainital Uttarakhand India Dept. of Computer Science and Engineering Dr. Akhilesh Das Gupta Institute of Technology and Management New Delhi India Faculty of Computer Science and Informatics Amman Arab University Mubis-Amman Jordan

Optimization of share price in order to evaluate whether to invest in specific share or not is subject of study paper. Optimized value has been utilized to enhance the performance and accuracy when dataset is trained using CNN. Review has been given on big cap, mid cap & small cap fund which are accessible in large, medium and small amount for the purpose of investing. It has been found that share in big cap are more dependable but offers limited return. However share in case of mid cap are considered comparatively more hazardous than big cap yet return may be substantial. The suggested study has given mechanism that will make use of optimization mechanism to filter the data set considering PSO and MVO mechanism. The suggested CNN based training model is claimed high performance system with improved accuracy.

关键词：

来源：评论

学校读者我要写书评

暂无评论

***: A Comprehensive Genomic Visualisation database for Malaysian Species

***: A Comprehensive Genomic Visualisation Database for Mala...

引用

Computing (ICOCO), IEEE International Conference on

作者： Vei Siang Chan Farhan Bin Mohamed Faezah Mohd Salleh Andres Iglesias Alina Binti Wagiran Ali Bin Selamat School of Computing Faculty of Engineering Universiti Teknologi Malaysia Johor Bahru Malaysia Media and Game Innovation Center of Excellence Universiti Teknologi Malaysia Johor Bahru Malaysia Dept. of Biosciences Faculty of Science Universiti Teknologi Malaysia Johor Bahru Malaysia Dept. of Applied Mathematics and Computational Sciences University of Cantabria Santander Spain Malaysia-Japan International Institute of Technology University of Technology Malaysia Kuala Lumpur Malaysia

ISBN: (纸本)9781665489973

Malaysia is one of the most megadiverse regions in the world. Therefore, it is important to record these rich biodiversity data to safeguard and improve national conservation efforts. However, the lack of nationwide genomic sequencing hampers a more sophisticated species identification, data management, and analytics activities. Thus, this project aims to design and implement ***, a comprehensive one-stop online DNA reference database with integrated DNA visualisation and digital references to allow the users to search, retrieve and analyse the collected DNA sequencing data interactively. This paper documented the system architecture and implementation methods to create ***. The results of *** can display taxonomy information of biodiversity in the web’s user interfaces and visualise the complex set of DNA data to allow new insights. Lastly, the authors identified several future works to improve *** including improve user interaction, add visualisation methods to support more bioinformatics tasks, and conduct formal user evaluation.

关键词： Sequential analysis Annotations Taxonomy data visualization DNA Genomics Systems architecture

来源：评论

学校读者我要写书评

暂无评论

Learning and planning for time-varying MDPs using maximum likelihood estimation

The Journal of Machine Learning Research

引用

The Journal of Machine Learning Research 2021年第1期22卷 1656-1695页

作者： Melkior Ornik Ufuk Topcu Department of Aerospace Engineering and the Coordinated Science Laboratory University of Illinois at Urbana-Champaign Urbana IL Dept. of Aero. Eng. and Eng. Mechanics and the Oden Inst. for Computational Eng. and Sciences University of Texas at Austin Austin TX

This paper proposes a formal approach to online learning and planning for agents operating in a priori unknown, time-varying environments. The proposed method computes the maximally likely model of the environment, given the observations about the environment made by an agent earlier in the system run and assuming knowledge of a bound on the maximal rate of change of system dynamics. Such an approach generalizes the estimation method commonly used in learning algorithms for unknown Markov decision processes with time-invariant transition probabilities, but is also able to quickly and correctly identify the system dynamics following a change. Based on the proposed method, we generalize the exploration bonuses used in learning for time-invariant Markov decision processes by introducing a notion of uncertainty in a learned time-varying model, and develop a control policy for time-varying Markov decision processes based on the exploitation and exploration trade-off. We demonstrate the proposed methods on four numerical examples: a patrolling task with a change in system dynamics, a two-state MDP with periodically changing outcomes of actions, a wind ow estimation task, and a multi-armed bandit problem with periodically changing probabilities of different rewards.

关键词： Markov decision processes changing environment maximum likelihood estimation online learning uncertainty quantification

来源：评论

学校读者我要写书评

暂无评论

Predictive Resilience Framework for Energy Management Systems in Next-Generation Smart Grids Using a Hybrid LSTM-ARIMA Model

Predictive Resilience Framework for Energy Management System...

引用

Innovations in High Speed Communication and Signal Processing (IHCSP), International Conference on

作者： Divya Nimma Shamim Ahmad Khan Gunjan Chhabra Prasanta Parida Ajmeera Kiran B Kiran Bala University of Southern Mississippi Data Analyst in UMMC USA Glocal School of Science & Technology Glocal University U.P India Department of CSE Graphic Era Hill University Dehradun Graphic Era Deemed to be University Dehradun Uttarakhand India School of Rural Management KIIT University Bhubaneswar Odisha India Dept. of CSE MLR Institute of Technology Hyderabad Telangana India Department of AI & DS K.Ramakrishnan College of Engineering Trichy India

ISBN: (数字)9798350368949

ISBN: (纸本)9798350368956

The Predictive Resilience Framework for Energy Management Systems in Next-Generation Smart Grids integrates advanced forecasting and modeling techniques to enhance grid stability and efficiency. This framework uses Long Short Term Memory (LSTM) network with AutoRegressive Integrated Moving Average (ARIMA) models as the backbone to make the accurate predictions and understanding the dynamics of the energy system. The use of ARIMA befits a number of time series forecasting requirements, which consists of identifying patterns based on past data, and mitigating non-stationary beliefs as load forecasting, renewable integration, fault identification, and demand response. On the other hand, LSTM networks are effective at modelling the long-term dependencies in the sequences hence enhancing the way the model produces future state from previous sequences. The hybrid LSTM-ARIMA approach merges the strengths of both models: ARIMA gives a stable framework on which to build the prediction and LSTM gives an improvement on these predictions due to its ability to deal with complex patterns and temporal attributes. This combination also helps fundamental predictive resilience by allowing for prevision of both energy demand and renewable production, standout of system abnormalities, and direction of peak loads and response programmes. Furthermore, the framework helps in analysing the ability of the grid to operate under various stress conditions such as weather conditions and cyber attacks. By including such superior modelling approaches the framework has been designed to enhance the utility of the grid, balance the energy provision and strengthen the general resource of the system to new threats and instability within the future and more complex smart grid systems.

关键词： Fault diagnosis Renewable energy sources Predictive models Smart grids Forecasting Energy management systems Long short term memory Next generation networking Load modeling Resilience

来源：评论

学校读者我要写书评

暂无评论

Building segmentation from airborne vhr images using mask r-cnn 4

Building segmentation from airborne vhr images using mask r-...

引用

4th ISPRS Geospatial Week 2019

作者： Zhou, K. Chen, Y. Smal, I. Lindenbergh, R. Dept. of Geoscience and Remote Sensing Delft University of Technology Netherlands Dept. of Computational Science and Engineering Delft University of Technology Netherlands

Up-to-date 3D building models are important for many applications. Airborne very high resolution (VHR) images often acquired annually give an opportunity to create an up-to-date 3D model. Building segmentation is often the first and utmost step. Convolutional neural networks (CNNs) draw lots of attention in interpreting VHR images as they can learn very effective features for very complex scenes. This paper employs Mask R-CNN to address two problems in building segmentation: detecting different scales of building and segmenting buildings to have accurately segmented edges. Mask R-CNN starts from feature pyramid network (FPN) to create different scales of semantically rich features. FPN is integrated with region proposal network (RPN) to generate objects with various scales with the corresponding optimal scale of features. The features with high and low levels of information are further used for better object classification of small objects and for mask prediction of edges. The method is tested on ISPRS benchmark dataset by comparing results with the fully convolutional networks (FCN), which merge high and low level features by a skip-layer to create a single feature for semantic segmentation. The results show that Mask R-CNN outperforms FCN with around 15% in detecting objects, especially in detecting small objects. Moreover, Mask R-CNN has much better results in edge region than FCN. The results also show that choosing the range of anchor scales in Mask R-CNN is a critical factor in segmenting different scale of objects. This paper provides an insight into how a good anchor scale for different dataset should be chosen. © Authors 2019.

关键词： Buildings

来源：评论

学校读者我要写书评

暂无评论

A new smart-cropping pipeline for prostate segmentation using deep learning networks

arXiv

引用

arXiv 2021年

作者： Zaridis, Dimitrios G. Mylona, Eugenia Tachos, Nikolaos S. Marias, Kostas Papanikolaou, Nikolaos Tsiknakis, Manolis Fotiadis, Dimitrios I. The Dept. of Biomedical Research FORTH-IMBB Ioannina Greece The Unit of Medical Technology and Intelligent Information Systems Dept. of Materials Science and Engineering University of Ioannina IoanninaGR 45110 Greece The Institute of Computer Science FORTH Heraklion Greece The Dept. of Electrical and Computer Engineering Hellenic Mediterranean University Greece The Centre for the Unknown Champalimaud Foundation Clinical Computational Imaging Group Lisbon Portugal Crete Heraklion Greece The Dept. of Biomedical Research Institute of Molecular Biology and Biotechnology FORTH Ioannina Greece The Dept. of Materials Science and Engineering Unit of Medical Technology and Intelligent Information Systems University of Ioannina IoanninaGR 45110 Greece

Prostate segmentation from magnetic resonance imaging (MRI) is a challenging task. In recent years, several network architectures have been proposed to automate this process and alleviate the burden of manual annotation. Although the performance of these models has achieved promising results, there is still room for improvement before these models can be used safely and effectively in clinical practice. One of the major challenges in prostate MR image segmentation is the presence of class imbalance in the image labels where the background pixels dominate over the prostate. In the present work we propose a DL-based pipeline for cropping the region around the prostate from MRI images to produce a more balanced distribution of the foreground pixels (prostate) and the background pixels and improve segmentation accuracy. The effect of DL-cropping for improving the segmentation performance compared to standard center-cropping is assessed using five popular DL networks for prostate segmentation, namely U-net, U-net+, Res Unet++, Bridge U-net and Dense U-net. The proposed smart-cropping outperformed the standard center cropping in terms of segmentation accuracy for all the evaluated prostate segmentation networks. In terms of Dice score, the highest improvement was achieved for the Unet+ and ResU-net++ architectures corresponding to 8.9% and 8%, respectively. © 2021, CC BY.

关键词： Image segmentation

来源：评论

学校读者我要写书评

暂无评论

A convolutional LSTM based residual network for deepfake video detection

arXiv

引用

arXiv 2020年

作者： Tariq, Shahroz Lee, Sangyup Woo, Simon S. Dept. of Computer Science and Engineering Sungkyunkwan University Suwon Korea Republic of Dept. of Applied Data Science Sungkyunkwan University Suwon Korea Republic of

In recent years, deep learning-based video manipulation methods have become widely accessible to masses. With little to no effort, people can easily learn how to generate deepfake videos with only a few victims or target images. This creates a significant social problem for everyone whose photos are publicly available on the Internet, especially on social media websites. Several deep learning-based detection methods have been developed to identify these deepfakes. However, these methods lack generalizability, because they perform well only for a specific type of deepfake method. Therefore, those methods are not transferable to detect other deepfake methods. Also, they do not take advantage of the temporal information of the video. In this paper, we addressed these limitations. We developed a Convolutional LSTM based Residual Network (CLRNet), which takes a sequence of consecutive images as an input from a video to learn the temporal information that helps in detecting unnatural looking artifacts that are present between frames of deepfake videos. We also propose a transfer learning-based approach to generalize different deepfake methods. Through rigorous experimentations using the FaceForensics++ dataset, we showed that our method outperforms five of the previously proposed state-of-the-art deepfake detection methods by better generalizing at detecting different deepfake methods using the same model. Copyright © 2020, The Authors. All rights reserved.

关键词： Convolution

来源：评论

学校读者我要写书评

暂无评论

Evaluating deep transfer learning for whole-brain cognitive decoding

arXiv

引用

arXiv 2021年

作者： Thomas, Armin W. Lindenberger, Ulman Samek, Wojciech Müller, Klaus-Robert Machine Learning Group Dept. of Computer Science and Electrical Engineering Technische Universität Berlin Berlin Germany Center for Lifespan Psychology Max Planck Institute for Human Development Berlin Germany Stanford Data Science Stanford University StanfordCA United States Dept. of Psychology Stanford University StanfordCA United States Max Planck UCL Centre for Computational Psychiatry and Ageing Research Berlin Germany Dept. of Artificial Intelligence Fraunhofer Heinrich Hertz Institute Berlin Germany BIFOLD – Berlin Institute for the Foundations of Learning and Data Berlin Germany Dept. of Artificial Intelligence Korea University Seoul Korea Republic of Max Planck Institute for Informatics Saarbrücken Germany

Research in many fields has shown that transfer learning (TL) is well-suited to improve the performance of deep learning (DL) models in datasets with small numbers of samples. This empirical success has triggered interest in the application of TL to cognitive decoding analyses with functional neuroimaging data. Here, we systematically evaluate TL for the application of DL models to the decoding of cognitive states (e.g., viewing images of faces or houses) from whole-brain functional Magnetic Resonance Imaging (fMRI) data. We first pre-train two DL architectures on a large, public fMRI dataset and subsequently evaluate their performance in an independent experimental task and a fully independent dataset. The pre-trained models consistently achieve higher decoding accuracies and generally require less training time and data than model variants that were not pre-trained, clearly underlining the benefits of pre-training. We demonstrate that these benefits arise from the ability of the pre-trained models to reuse many of their learned features when training with new data, providing deeper insights into the mechanisms giving rise to the benefits of pre-training. Yet, we also surface nuanced challenges for whole-brain cognitive decoding with DL models when interpreting the decoding decisions of the pre-trained models, as these have learned to utilize the fMRI data in unforeseen and counterintuitive ways to identify individual cognitive states. © 2021, CC BY.

关键词： Decoding

来源：评论

学校读者我要写书评

暂无评论

Validation free and replication robust volume-based data valuation 21

Validation free and replication robust volume-based data val...

引用

Proceedings of the 35th International Conference on Neural Information Processing Systems

作者： Xinyi Xu Zhaoxuan Wu Chuan Sheng Foo Bryan Kian Hsiang Low Dept. of Computer Science National University of Singapore Republic of Singapore and Institute for Infocomm Research A*STAR Republic of Singapore Institute of Data Science National University of Singapore Republic of Singapore and Integrative Sciences and Engineering Programme NUSGS Republic of Singapore Institute for Infocomm Research A*STAR Republic of Singapore Dept. of Computer Science National University of Singapore Republic of Singapore

ISBN: (纸本)9781713845393

data valuation arises as a non-trivial challenge in real-world use cases such as collaborative machine learning, federated learning, trusted data sharing, data marketplaces. The value of data is often associated with the learning performance (e.g., validation accuracy) of a model trained on the data, which introduces a close coupling between data valuation and validation. However, a validation set may not be available in practice and it can be challenging for the data providers to reach an agreement on the choice of the validation set. Another practical issue is that of data replication: Given the value of some data points, a dishonest data provider may replicate these data points to exploit the valuation for a larger reward/payment. We observe that the diversity of the data points is an inherent property of a dataset that is independent of validation. We formalize diversity via the volume of the data matrix (i.e., determinant of its left Gram), which allows us to establish a formal connection between the diversity of data and learning performance without requiring validation. Furthermore, we propose a robust volume measure with a theoretical guarantee on the replication robustness by following the intuition that copying the same data points does not increase the diversity of data. We perform extensive experiments to demonstrate its consistency in valuation and practical advantages over existing baselines and show that our method is model- and task-agnostic and can be flexibly adapted to handle various neural networks.

关键词：

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：