检索结果-内蒙古大学图书馆

Methods for Predictive Performance Improvement of Deep Learning Systems for Aerial Building Roof Detection with Multispectral Images

Methods for Predictive Performance Improvement of Deep Learn...

引用

IEEE International conference on Computational Intelligence and Virtual Environments for Measurement Systems and Applications

作者： A. Churi D. B. Megherbi Department of Electrical and Computer Engineering Center for Human/Machine Intelligence Networking and Distributed Systems University of Massachusetts Lowell MA USA

In this paper, we present a dilated convolutional neural network-based deep learning system for the challenging problem of building accurate roof pixel detection from city aerial multispectral images in the visible and infrared spectral bands. The proposed system uses an automatic spectral band selection strategy and an automatic histogram-based threshold-finding method to improve the network's prediction accuracy performance. Current existing systems train a single model on all the cities of interest data at once. These systems also usually use only RGB bands and avoid post-processing of the predicted images for increased prediction accuracies. In this paper, we present the following solutions to these object detection challenges, 1) we propose and implement a separate model for each city dataset, 2) we propose an optimal band selection strategy to choose the most informative visible-infrared bands from multispectral images, 3) we propose an adaptive and automatic histogram-based optimal threshold finding technique for building roof object segmentation. Our results show that having a separate model for each city rather than only one model for all the combined cities increases the system's performance by 20.77 % in building roof pixel prediction in any of the cities considered. We show that our system also has a 35.6% increase in F1 score compared to its state-of-the-art counterpart, the winning implementation of the SpaceNet building detection national challenge, which took place in 2017.

关键词：

来源：评论

学校读者我要写书评

暂无评论

SFTA: Spiking neural networks Vulnerable to Spiking Feature Transferable Attack

SFTA: Spiking Neural Networks Vulnerable to Spiking Feature ...

引用

IEEE International Symposium on Parallel and distributed processing with Applications and IEEE International conference on Ubiquitous Computing and Communications (ISPA/IUCC)

作者： Xuanwei Lin Chen Dong Ximeng Liu College of Computer and Data Science Fuzhou University Fujian Key Laboratory of Network Computing and Intelligent Information Processing (Fuzhou University) Key Lab of Information Security of Network Systems (Fuzhou University)

Many recent works have shown that spiking neural networks (SNNs) are vulnerable to well-crafted adversarial attacks in the white-box setting, which may lead to misclassification of the trained model. However, the information about the model is in a black-box setting and is usually unknown to the attacker, e.g., the parameters and structure of the model. So it is more challenging to counter the attack in the black-box setting, and there is still little research on this aspect in SNNs. Therefore, this work describes a transfer-based black box adversarial attack against spike feature space representations, providing insights for robust spike features in SNNs models. The attack is explicitly designed to be transferable, and more robust feature gradients are obtained by accumulating the feature gradients of the intermediate layers. Then we can successfully attack the black-box model by suppressing robust positive features and enhancing negative features. Extensive experimental evaluation demonstrates the proposed method’s effectiveness and superior performance on static and neuromorphic datasets. i.e., The attack success is 89.72% and 90% on the static MNIST and CIFAR10 datasets, and 85.97% and 69.09% on the neuromorphic N-MNIST and CIFARO-DVS datasets.

关键词： Correlation Neuromorphics neural networks Closed box Ubiquitous computing Optimization Glass box

来源：评论

学校读者我要写书评

暂无评论

distributed representation and one-hot representation fusion with gated network for clinical semantic textual similarity

引用

BMC MEDICAL INFORMATICS AND DECISION MAKING 2020年第1-Sup期20卷 72-72页

作者： Xiong, Ying Chen, Shuai Qin, Haoming Cao, He Shen, Yedan Wang, Xiaolong Chen, Qingcai Yan, Jun Tang, Buzhou Harbin Inst Technol Dept Comp Sci Shenzhen Guangdong Peoples R China Peng Cheng Lab Shenzhen Guangdong Peoples R China Yidu Cloud Beijing Technol Co Ltd Beijing Peoples R China

Background Semantic textual similarity (STS) is a fundamental natural language processing (NLP) task which can be widely used in many NLP applications such as Question Answer (QA), Information Retrieval (IR), etc. It is a typical regression problem, and almost all STS systems either use distributed representation or one-hot representation to model sentence pairs. Methods In this paper, we proposed a novel framework based on a gated network to fuse distributed representation and one-hot representation of sentence pairs. Some current state-of-the-art distributed representation methods, including Convolutional neural network (CNN), Bi-directional Long Short Term Memory networks (Bi-LSTM) and Bidirectional Encoder Representations from Transformers (BERT), were used in our framework, and a system based on this framework was developed for a shared task regarding clinical STS organized by BioCreative/OHNLP in 2018. Results Compared with the systems only using distributed representation or one-hot representation, our method achieved much higher Pearson correlation. Among all distributed representations, BERT performed best. The highest Person correlation of our system was 0.8541, higher than the best official one of the BioCreative/OHNLP clinical STS shared task in 2018 (0.8328) by 0.0213. Conclusions distributed representation and one-hot representation are complementary to each other and can be fused by gated network.

关键词： Clinical semantic textual similarity Gated network distributed representation One-hot representation

来源：评论

学校读者我要写书评

暂无评论

DeepHop on Edge: Hop-by-hop Routing by distributed Learning with Semantic Attention 20

DeepHop on Edge: Hop-by-hop Routing by Distributed Learning ...

引用

49th International conference on Parallel processing (ICPP)

作者： He, Bo Wang, Jingyu Qi, Qi Sun, Haifeng Zhuang, Zirui Liu, Cong Liao, Jianxin Beijing Univ Posts & Telecommun Beijing Peoples R China China Mobile Res Inst Beijing Peoples R China

ISBN: (纸本)9781450388160

Multi-access Edge Computing (MEC) and ubiquitous smart devices help serve end-users efficiently and optimally through providing emerging edge-deployed services. Meanwhile, heavy and time-varying traffic loads are produced in the edge network, so that an efficient traffic forwarding mechanism is required. In this paper, we propose a parallel and distributed learning approach, DeepHop, to adapt to the volatile environments and realize hop-by-hop routing. The Multi-Agent Deep Reinforcement Learning (MADRL) is used to alleviate the edge network congestion and maximize the utilization of network resources. DeepHop determines the routing among edge network nodes for heterogeneous types of traffic according to the current workload and capability. By joining with an attention mechanism, DeepHop obtains the semantics from the elements of the network state to help the agents learn the importance of each element on routing. Experiment results show that DeepHop achieves the increase of successfully transmitted packets by 15% compared with the state-of-the-art algorithms. Besides, DeepHop with an attention mechanism reduces convergence time by nearly half compared with the common-used structures of neural networks.

关键词： distributed routing wireless edge networks multi-agent reinforcement learning self-attention

来源：评论

学校读者我要写书评

暂无评论

ArchGAN: A Generative Adversarial network for Architectural Distortion Abnormalities in Digital Mammograms

ArchGAN: A Generative Adversarial Network for Architectural ...

引用

IEEE International conference on Electrical, Computer, and Energy Technologies (ICECET)

作者： Oyelade, Olaide N. Ezugwu, Absalom E. Univ KwaZulu Natal Sch Math Stat & Comp Sci Pietermaritzburg Campus ZA-3201 Pietermaritzburg Kwazulu Natal South Africa

ISBN: (纸本)9781665442312

Applying deep learning (DL) models to medical image processing has yielded impressive performances. Generative adversarial networks (GAN) are examples of DL models proposed to mediate the challenge of insufficient image data by synthesizing data similar to the real sample images. However, we observed that benchmark datasets with the region of interest (ROIs) for characterizing abnormalities in breast cancer using digital mammography have only limited image samples with architectural distortion abnormalities. Images with architectural distortion abnormality in digital mammograms are sparsely distributed across all publicly available datasets. This paper proposes a GAN model named ArchGAN, which is aimed at synthesizing ROI-based samples with architectural distortion abnormality for training convolutional neural networks (CNNs). Our approach involves the design of the GAN model, consisting of both generator and discriminator, to learn a hierarchy of representation in digital mammograms. The proposed GAN model was applied to Mammographic Image Analysis Society (MIAS) and Digital Database for Screening Mammography (DDSM) datasets. In addition, the quality of images generated was also evaluated using PSNR, SSIM, FSIM, BRISQUE, PQUE, NIQUE, and FID. Results obtained showed that ArchGAN performed very well. The outcome of this study is a model for augmenting CNN models with ROI-centric images with architectural distortion abnormalities in digital mammography.

关键词： Generative adversarial networks generator discriminator

来源：评论

学校读者我要写书评

暂无评论

Dynamic QoS Prediction Based on Attention Mechanism and Recurrent neural network

Dynamic QoS Prediction Based on Attention Mechanism and Recu...

引用

IEEE International conference on Big Data and Cloud Computing (BdCloud)

作者： Yingxue Wang Qin Lu Yichao Wang Mengwei Wu Weixiao Li School of Computer Science and Technology Qilu University of Technology (Shandong Academy of Sciences) Jinan China Information Technology Center China Mobile Communication Group Co. Ltd Beijing China

With the growth of science and technology, there are more and more services in the network, and their functions similar. It has become crucial to figure out how to recommend real-time personalized services based on user and service history. Quality of Service (QoS), as a noncritical attribute of Web services, it is a vital criterion to measure the usage performance and utility of service. The dynamic prediction of QoS is a key factor in the recommendation process. In this paper, a new time-aware QoS prediction model is proposed. It extracte global features while considering its dynamics, it capture the implicit features within a time period using Convolutional neural networks(CNN) and Attention Mechanisms. The model not only takes into account the impact of time, but also better learns the implicit information between users and services during the time interval, forming global features within time. It study QoS at various times from the entire time period, capture key features, thereby make dynamic QoS prediction. The proposed model has better results by extensive experiments on real dataset.

关键词： Recurrent neural networks Web services Quality of service Predictive models Feature extraction Real-time systems Data mining

来源：评论

学校读者我要写书评

暂无评论

distributed online convex optimization with compressed communication 22

Distributed online convex optimization with compressed commu...

引用

Proceedings of the 36th International conference on neural Information processing Systems

作者： Zhipeng Tu Xi Wang Yiguang Hong Lei Wang Deming Yuan Guodong Shi Academy of Mathematics and Systems Science Chinese Academy of Sciences China and Australian Center for Field Robotics School of AMME The University of Sydney Australia Department of Control Science and Engineering Tongji University China College of Control Science and Engineering Zhejiang University China School of Automation Nanjing University of Science and Technology China Australian Center for Field Robotics School of AMME The University of Sydney Australia

ISBN: (纸本)9781713871088

We consider a distributed online convex optimization problem when streaming data are distributed among computing agents over a connected communication network. Since the data are high-dimensional or the network is large-scale, communication load can be a bottleneck for the efficiency of distributed algorithms. To tackle this bottleneck, we apply the state-of-art data compression scheme to the fundamental GD-based distributed online algorithms. Three algorithms with difference-compressed communication are proposed for full information feedback (DC-DOGD), one-point bandit feedback (DC-DOBD), and two-point bandit feedback (DC-DO2BD), respectively. We obtain regret bounds explicitly in terms of time horizon, compression ratio, decision dimension, agent number, and network parameters. Our algorithms are proved to be no-regret and match the same regret bounds, w.r.t. time horizon, with their uncompressed versions for both convex and strongly convex losses. Numerical experiments are given to validate the theoretical findings and illustrate that the proposed algorithms can effectively reduce the total transmitted bits for distributed online training compared with the uncompressed baseline.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Analyzing the distributed training of deep-learning models via data locality

Analyzing the distributed training of deep-learning models v...

引用

Euromicro conference on Parallel, distributed and network-Based processing

作者： Saúl Alonso-Monsalve Alejandro Calderón Félix García-Carballeira José Rivadeneira CERN Geneva Switzerland Universidad Carlos III de Madrid Leganés Madrid Spain

In the last few years, deep-learning models are becoming crucial for numerous scientific and industrial applications. Due to the growth and complexity of deep neural networks, researchers have been investigating techniques to train those networks more efficiently. Many efforts have been made to optimize deep-learning models by parallelizing or distributing their training computation across multiple devices. Current state-of-the-art techniques, such as Horovod, have shown to maximize the performance of both the training computation and the inter-node communication of models for different deep-learning frameworks. However, some applications cannot take advantage of the above techniques due to an I/O bottleneck caused by the input data, thus limiting the scalability of the trainings. In this paper, we study an approach based on data locality - that has not been fully studied yet - for those neural networks that cannot benefit from scaling their computation due to a significant bottleneck in the data I/O.

关键词： Training Performance evaluation Analytical models Limiting Computational modeling Scalability neural networks

来源：评论

学校读者我要写书评

暂无评论

Aperiodic Local SGD: Beyond Local SGD 22

Aperiodic Local SGD: Beyond Local SGD

引用

Proceedings of the 51st International conference on Parallel processing

作者： Hao Zhang Tingting Wu Siyao Cheng Jie Liu Harbin Institute of Technology China Harbin Institute of Technology (Shenzhen) China

ISBN: (纸本)9781450397339

Variations of stochastic gradient decedent (SGD) methods are at the core of training deep neural network models. However, in distributed deep learning, where multiple computing devices and data segments are employed in the training process, the performance of SGD can be significantly limited by the overhead of gradient communication. Local SGD methods are designed to overcome this bottleneck by averaging individual gradients trained over parallel workers after multiple local iterations. Currently, both for theoretical analyses and for practical applications, most studies employ periodic synchronization scheme by default, while few of them focus on the aperiodic schemes to obtain better performance models with limited computation and communication overhead. In this paper, we investigate local SGD with an arbitrary synchronization scheme to answer two questions: (1) Is the periodic synchronization scheme best? (2) If not, what is the optimal one? First, for any synchronization scheme, we derive the performance boundary with fixed overhead, and formulate the performance optimization under given computation and communication constraints. Then we find a succinct property of the optimal scheme that the local iteration number decreases as training continues, which indicates the periodic one is suboptimal. Furthermore, with some reasonable approximations, we obtain an explicit form of the optimal scheme and propose Aperiodic Local SGD (ALSGD) as an improved substitute for local SGD without any overhead increment. Our experiments also confirm that with the same computation and communication overhead, ALSGD outperforms local SGD in performance, especially for heterogeneous data.

关键词： computation and communication cost federated learning distributed learning local SGD data heterogeneity

来源：评论

学校读者我要写书评

暂无评论

distributed Accumulation based Energy Efficient STT-MRAM based Digital PIM Architecture

Distributed Accumulation based Energy Efficient STT-MRAM bas...

引用

International SoC Design conference, ISOCC

作者： Dongsu Kim Jongsun Park School of Electrical Engineering Korea University

ISBN: (纸本)9781665459723

Spin transfer torque MRAM (STT-MRAM) based digital processing-in-memory (PIM) has been recently proposed for energy-efficient processing of convolutional neural network (CNN) without analog-to-digital converters (ADC). However, since only computations between operands stored in the same row are possible in the digital PIM, it consumes considerable energy for data transfer between memory rows when computations between data stored in different rows are performed. In this paper, we present energy-efficient digital PIM architecture that supports the distributed accumulation scheme. In the proposed PIM architecture, the computations for accumulation are distributed to the memory arrays and peripheral circuits, and only a portion of the workload for the accumulations which does not require heavy data transfer cost is performed in the memory array. Then, the remaining of the accumulations that needs complicated data transfer among the memory array are performed using the peripheral circuits. In addition, as the value of the partial sums is read before accumulation, the zero-skipping technique, which cannot be applied when the computations are fully performed in the memory array, can be also applied in the proposed PIM architecture. The simulations with 28nm CMOS process show that the proposed digital PIM architecture with zero prediction achieves the energy savings up to 54.3% over the conventional digital PIMs.

关键词： Semiconductor device modeling Torque Simulation Predictive models Data transfer In-memory computing Energy efficiency

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：