检索结果-内蒙古大学图书馆

22nd IEEE International Symposium on Parallel and distributed processing with Applications, ISPA 2024

作者： You, Wei Zhao, Chunlei Shan, Chao Dong, Guozhong Yu, Jie Guan, Yong Tianjin University of Technology Tianjin Key Laboratory of Intelligent Computing and Software New Technology Tianjin China Coordination Center of China Hubei Branch of National Computer Network Emergency Response Technical Team HuBei China Pengcheng Laboratory Department of New Networks Shenzhen China

ISBN: (纸本)9798331509712

An increasing number of Advanced Persistent Threat (APT) organizations are utilizing DNS (Domain Name System) covert channels to evade network intrusion detection systems and establish private authoritative servers for remote attacks. However, current detection models for DNS covert channels often have poor performance and limited generalization ability. To address these challenges, we propose a DNS covert channel attack detection method based on spatiotemporal feature fusion. This method combines the Interaction timing sequence features of DNS traffic with the spatial statistical features of domain names, and constructs a detection model called ConvSLSTM, which integrates a Convolutional neural network (CNN) with a stacked Long Short Term Memory (LSTM) network. CNN extracts local spatial features from traffic, while stacked LSTM captures Interaction timing sequence features, significantly improving detection accuracy. This article uses three typical public datasets for experiments, which contain a total of 440000 DNS covert channel attack data. Comparative experiments show that our method is significantly superior to existing methods and exhibits stronger generalization ability in detecting DNS covert channel attacks constructed by unknown tools. © 2024 IEEE.

关键词： network intrusion

来源：评论

学校读者我要写书评

暂无评论

Perceptual Model Hashing: Towards neural network Model Authentication 24

Perceptual Model Hashing: Towards Neural Network Model Authe...

引用

IEEE 24th International Workshop on Multimedia Signal processing (MMSP)

作者： Li, Xinran Wang, Zichi Feng, Guorui Zhang, Xinpeng Qin, Chuan Univ Shanghai Sci & Technol Sch Opt Elect & Comp Engn Shanghai 200093 Peoples R China Shanghai Univ Sch Commun & Informat Engn Shanghai 200444 Peoples R China

ISBN: (数字)9781665471893

ISBN: (纸本)9781665471893

A lot of excellent neural network models are valuable wealth to the field of artificial intelligence, which may be plagiarized and distributed without authorization. For this reason, few research establishments and industries reveal the internals of their neural network models. To authenticate suspicious pirated models, this letter proposes a gray-box hashing method for the neural network models that designed for image classification. In the proposed method, the hash sequence of original model can be extracted without knowing both the structure and weight parameters except the vectors in output layer. To the best of our knowledge, this is the first work focusing on gray-box perceptual model hashing to identify and authenticate neural network models. Experimental results show that our method performs satisfactory perceptual robustness and discrimination capability, and can effectively classify perceptual similar versions of the original model and distinct models.

关键词： Model authentication neural network Model hashing Gray-box Adversarial examples

来源：评论

学校读者我要写书评

暂无评论

Efficient distributed Inference of Deep neural networks via Restructuring and Pruning 37

Efficient Distributed Inference of Deep Neural Networks via ...

引用

37th AAAI Conference on Artificial Intelligence (AAAI) / 35th Conference on Innovative Applications of Artificial Intelligence / 13th Symposium on Educational Advances in Artificial Intelligence

作者： Abdi, Afshin Rashidi, Saeed Fekri, Faramarz Krishna, Tushar Georgia Inst Technol Sch Elect & Comp Engn Atlanta GA 30332 USA

ISBN: (纸本)9781577358800

In this paper, we consider the parallel implementation of an already-trained deep model on multiple processing nodes (a.k.a. workers). Specifically, we investigate as to how a deep model should be divided into several parallel sub-models, each of which is executed efficiently by a worker. Since latency due to synchronization and data transfer among workers negatively impacts the performance of the parallel implementation, it is desirable to have minimum interdependency among parallel sub-models. To achieve this goal, we propose to rearrange the neurons in the neural network, partition them (without changing the general topology of the neural network), and modify the weights such that the interdependency among sub-models is minimized under the computations and communications constraints of the workers while minimizing its impact on the performance of the model. We propose RePurpose, a layer-wise model restructuring and pruning technique that guaran-tees the performance of the overall parallelized model. To efficiently apply RePurpose, we propose an approach based on l(0) optimization and the Munkres assignment algorithm. We show that, compared to the existing methods, RePurpose significantly improves the efficiency of the distributed inference via parallel implementation, both in terms of communication and computational complexity.

关键词： Data transfer

来源：评论

学校读者我要写书评

暂无评论

An on-chip photonic deep neural network for image classification

引用

NATURE 2022年第7914期606卷 501-+页

作者： Ashtiani, Farshid Geers, Alexander J. Aflatouni, Firooz Univ Penn Dept Elect & Syst Engn Philadelphia PA 19104 USA

Deep neural networks with applications from computer vision to medical diagnosis(1-5) are commonly implemented using clock-based processors(6-14), in which computation speed is mainly limited by the clock frequency and the memory access time. In the optical domain, despite advances in photonic computation(15-17), the lack of scalable on-chip optical non-linearity and the loss of photonic devices limit the scalability of optical deep networks. Here we report an integrated end-to-end photonic deep neural network (PDNN) that performs sub-nanosecond image classification through direct processing of the optical waves impinging on the on-chip pixel array as they propagate through layers of neurons. In each neuron, linear computation is performed optically and the non-linear activation function is realized opto-electronically, allowing a classification time of under 570 ps, which is comparable with a single clock cycle of state-of-the-art digital platforms. A uniformly distributed supply light provides the same per-neuron optical output range, allowing scalability to large-scale PDNNs. Two-class and four-class classification of handwritten letters with accuracies higher than 93.8% and 89.8%, respectively, is demonstrated. Direct, clock-less processing of optical data eliminates analogue-to-digital conversion and the requirement for a large memory module, allowing faster and more energy efficient neural networks for the next generations of deep learning systems.

关键词： Electrical and electronic engineering Optics and photonics Silicon photonics

来源：评论

学校读者我要写书评

暂无评论

Communication-Censored distributed Stochastic Gradient Descent

引用

IEEE TRANSACTIONS ON neural networkS AND LEARNING SYSTEMS 2022年第11期33卷 6831-6843页

作者： Li, Weiyu Wu, Zhaoxian Chen, Tianyi Li, Liping Ling, Qing Sun Yat Sen Univ Sch Comp Sci & Engn Guangzhou 510006 Peoples R China Univ Sci & Technol China Sch Gifted Young Hefei 230026 Peoples R China Sun Yat Sen Univ Guangdong Prov Key Lab Computat Sci Guangzhou 510006 Peoples R China Pazhou Lab Guangzhou 510300 Peoples R China Rensselaer Polytech Inst Dept Elect Comp & Syst Engn Troy NY 12180 USA Univ Sci & Technol China Dept Automat Hefei 230027 Peoples R China

This article develops a communication-efficient algorithm to solve the stochastic optimization problem defined over a distributed network, aiming at reducing the burdensome communication in applications, such as distributed machine learning. Different from the existing works based on quantization and sparsification, we introduce a communication-censoring technique to reduce the transmissions of variables, which leads to our communication-censored distributed stochastic gradient descent (CSGD) algorithm. Specifically, in CSGD, the latest minibatch stochastic gradient at a worker will be transmitted to the server if and only if it is sufficiently informative. When the latest gradient is not available, the stale one will be reused at the server. To implement this communication-censoring strategy, the batch size is increasing in order to alleviate the effect of stochastic gradient noise. Theoretically, CSGD enjoys the same order of convergence rate as that of SGD but effectively reduces communication. Numerical experiments demonstrate the sizable communication saving of CSGD.

关键词： Servers Convergence Optimization Stochastic processes Machine learning algorithms Sun Signal processing algorithms Communication censoring communication efficiency distributed optimization stochastic gradient descent (SGD)

来源：评论

学校读者我要写书评

暂无评论

AutoPipe: Automatic Configuration of Pipeline Parallelism in Shared GPU Cluster 24

AutoPipe: Automatic Configuration of Pipeline Parallelism in...

引用

53rd International Conference on Parallel processing (ICPP)

作者： Hu, Jinbin Wang, Hao Liu, Ying Wang, Jin Changsha Univ Sci & Technol Changsha Peoples R China Hong Kong Univ Sci & Technol Hong Kong Peoples R China Hunan Univ Sci & Technol Xiangtan Peoples R China

ISBN: (纸本)9798400717932

As training Deep neural network (DNN) is time-consuming, people resort to parallelization across multiple accelerators. A plethora of solutions adopt data/model parallelization, but they suffer from frequent-weight synchronization overhead or resource under-utilization. Recent work introduces pipeline parallelism to improve the utilization of accelerators, however, most existing pipeline parallelism approaches take a one-shot configuration, while ignoring the fluctuation of available resources, e.g., bandwidth and GPUs. Moreover, the heuristic work partition methods oversimplify the computation and communication process, leading to sub-optimal results. To address this challenge, we present AutoPipe, a self-adaptive pipeline parallelism optimization solution. At its core, AutoPipe introduces a reinforcement learning (RL) based work partitioning model, which takes into account both exact communication procedure and dynamic state switching. To mitigate the stalls on state switching, AutoPipe adopts layer-by-layer computation under switching. We have implemented an AutoPipe prototype and evaluated it via testbed experiments. Our results show that the AutoPipe-enhanced PipeDream can find better work partitioning and benefit from dynamic configuration, outperforming the vanilla solutions by up to 89% for exclusive tasks and 143% in dynamic workloads. Furthermore, we show that AutoPipe can also work well with other pipeline parallelism schemes and achieve considerable performance gains.

关键词： distributed Training Deep neural network Pipeline Parallelism Reinforcement Learning

来源：评论

学校读者我要写书评

暂无评论

Dissociable default-mode subnetworks subserve childhood attention and cognitive flexibility: Evidence from deep learning and stereotactic electroencephalography

引用

neural networkS 2023年 167卷 827-837页

作者： Warsi, Nebras M. Wong, Simeon M. Germann, Juergen Boutet, Alexandre Arski, Olivia N. Anderson, Ryan Erdman, Lauren Yan, Han Suresh, Hrishikesh Gouveia, Flavia Venetucci Loh, Aaron Elias, Gavin J. B. Kerr, Elizabeth Smith, Mary Lou Ochi, Ayako Otsubo, Hiroshi Sharma, Roy Jain, Puneet Donner, Elizabeth Lozano, Andres M. Snead, O. Carter Ibrahim, George M. Hosp Sick Children Div Neurosurg 555 Univ Ave Toronto ON Canada Univ Toronto Dept Biomed Engn Toronto ON Canada Hosp Sick Children Program Neurosci & Mental Hlth Toronto ON Canada Univ Hlth Network Toronto Western Hosp Div Neurosurg Toronto ON Canada Univ Toronto Joint Dept Med Imaging Toronto ON Canada Maerospace Corp Waterloo ON Canada Univ Hlth Network Vector Inst Artificial Intelligence Toronto ON Canada Univ Toronto Hosp Sick Children Dept Psychol 555 Univ Ave Toronto ON M5G 1X8 Canada Hosp Sick Children Abe Bresver Chair Funct Neurosurg Div Neurosurg 555 Univ AveRoom 1503 Toronto ON M5G 1X8 Canada

Cognitive flexibility encompasses the ability to efficiently shift focus and forms a critical component of goal-directed attention. The neural substrates of this process are incompletely understood in part due to difficulties in sampling the involved circuitry. We leverage stereotactic intracranial recordings to directly resolve local-field potentials from otherwise inaccessible structures to study moment to-moment attentional activity in children with epilepsy performing a flexible attentional task. On an individual subject level, we employed deep learning to decode neural features predictive of task performance indexed by single-trial reaction time. These models were subsequently aggregated across participants to identify predictive brain regions based on AAL atlas and FIND functional network parcellations. Through this approach, we show that fluctuations in beta (12-30 Hz) and gamma (30-80 Hz) power reflective of increased top-down attentional control and local neuronal processing within relevant large-scale networks can accurately predict single-trial task performance. We next performed connectomic profiling of these highly predictive nodes to examine task-related engagement of distributed functional networks, revealing exclusive recruitment of the dorsal default mode network during shifts in attention. The identification of distinct substreams within the default mode system supports a key role for this network in cognitive flexibility and attention in children. Furthermore, convergence of our results onto consistent functional networks despite significant inter subject variability in electrode implantations supports a broader role for deep learning applied to intracranial electrodes in the study of human attention.(c) 2023 Elsevier Ltd. All rights reserved.

关键词： Deep learning ADHD Attention Shapley values Default mode network

来源：评论

学校读者我要写书评

暂无评论

Spatio-Temporal Graph neural networks for Multi-Site PV Power Forecasting

引用

IEEE TRANSACTIONS ON SUSTAINABLE ENERGY 2022年第2期13卷 1210-1220页

作者： Simeunovic, Jelena Schubnel, Baptiste Alet, Pierre-Jean Carrillo, Rafael E. CSEM PV Ctr CH-2002 Neuchatel Switzerland CSEM CH-2002 Neuchatel Switzerland Ecole Polytech Fed Lausanne Signal Proc Lab LTS4 CH-1015 Lausanne Switzerland

Accurate forecasting of solar power generation with fine temporal and spatial resolution is vital for the operation of the power grid. However, state-of-the-art approaches that combine machinelearning with numerical weather predictions (NWP) have coarse resolution. In this paper, we take a graph signal processing perspective and model multi-site photovoltaic (PV) production time series as signals on a graph to capture their spatio-temporal dependencies and achieve higher spatial and temporal resolution forecasts. We present two novel graph neural network models for deterministic multi-site PV forecasting dubbed the graph-convolutional long short term memory (GCLSTM) and the graph-convolutional transformer (GCTrafo) models. These methods rely solely on production data and exploit the intuition that PV systems provide a dense network of virtual weather stations. The proposed methods were evaluated in two data sets for an entire year: 1) production data from 304 real PV systems, and 2) simulated production of 1000 PV systems, both distributed over Switzerland. The proposed models outperform state-of-the-art multi-site forecasting methods for prediction horizons of six hours ahead. Furthermore, the proposed models outperform state-of-the-art single-site methods with NWP as inputs on horizons up to four hours ahead.

关键词： Forecasting Convolution Predictive models Production Weather forecasting Correlation Data models Photovoltaic systems forecasting machine learning graph signal processing graph neural networks

来源：评论

学校读者我要写书评

暂无评论

LGDB-Net: Dual-Branch Path for Building Extraction from Remote Sensing Image 30

LGDB-Net: Dual-Branch Path for Building Extraction from Remo...

引用

30th IEEE International Conference on Parallel and distributed Systems, ICPADS 2024

作者： Zhang, Ronghuan Zhao, Jing Li, Ming Zou, Qingzhi Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center Jinan China Shandong Engineering Research Center of Big Data Applied Technology Faculty of Computer Science and Technology Jinan China Shandong Fundamental Research Center for Computer Science Shandong Provincial Key Laboratory of Computer Networks Jinan China Shandong University of Traditional Chinese Medicine School of Intelligence and Information Engineering Jinan China

ISBN: (纸本)9798331515966

Extracting buildings from remote sensing images using deep learning techniques is a widely applied and crucial task. Convolutional neural networks (CNNs) adopt hierarchical feature representation, showcasing powerful capabilities in extracting local information but facing challenges in capturing global features. Transformers can address this limitation, but they perform poorly in extracting local features and significantly increase memory requirements and computational complexity. To overcome these challenges, we propose a method for building extraction from remote sensing images called LGDB-Net (Local-Global Dual-Branch network), employing a dual-branch approach. Firstly, inspired by Swin Transformer, we designed GB-Former(Global Branch-Former) as the backbone network to model global information. We use a linear multi-head self-attention mechanism to reduce time and memory complexity while maintaining a large global receptive field. Additionally, we replace the traditional multi-layer perceptron with a convolution-enhanced multi-layer perceptron to improve channel feature representation, reduce model parameters, and enhance segmentation performance. Secondly, we use multiple Depth-wise Conv3×3 + LN (Layer Normalization) + GeLU (Gaussian Error Linear Unit) modules as the auxiliary branch for local detailed feature extraction. Finally, we adopt a multi-scale feature fusion strategy to integrate feature information from both branches. We conduct a series of experiments on three datasets: WHU, Massachusetts, and Inria. The experimental results demonstrate that the proposed method not only effectively improves segmentation accuracy with lower building omission and commission rates but also significantly reduces model parameters and computational complexity. © 2024 IEEE.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Automatic Mixed Precision and distributed Data-Parallel Training for Satellite Image Classification using CNN 15

Automatic Mixed Precision and Distributed Data-Parallel Trai...

引用

15th International Conference on Machine Vision (ICMV)

作者： Nuwara, Yohanes Kitt, Wong W. Juwono, Filbert H. Ollivierre, Gregory Asia Pulp & Paper Sinarmas Land Plaza MH Thamrin Jakarta Indonesia Curtin Univ Malaysia CDT 250 Bellevue WA 98009 USA Univ Southampton Malaysia Iskandar Puteri 79100 Johor Malaysia OmegaCrop 71-75 Shelton St London England

ISBN: (纸本)9781510666184;9781510666191

Deep learning models for computer vision in remote sensing such as Convolutional neural network (CNN) has benefitted acceleration from the usage of multiple CPUs and GPUs. There are several ways to make the training stage more effective in terms of utilizing multiple cores at the same time by processing different image mini-batches with a duplicated model called distributed Data Parallelization (DDP) and computing the parameters in a lower precision floating-point number called Automatic Mixed Precision (AMP). We would like to investigate the impact of DDP and AMP training modes on the overall utilization and memory consumption of CPU and GPU, as well as the accuracy of a CNN model. The study is performed on the EuroSAT dataset, a Sentinel-2-based benchmark satellite image dataset for image classification of land covers. We compare training using 1 CPU, using DDP, and using both DDP and AMP over 100 epochs using ResNet-18 architecture. The hardware that we used are Intel Xeon Silver 4116 CPU with 24 cores and an NVIDIA v100 GPU. We find that although parallelization of CPUs or DDP takes less time to train on the images, it can take 50 MB more memory than using only a single CPU. The combination of DDP and AMP can release memory up to 160 MB and reduce computation time by 20 seconds. The test accuracy is slightly higher for both DDP and DDP-AMP at 90.61% and 90.77% respectively than without DDP and AMP at 89.84%. Hence, training using distributed Data Parallelization (DDP) and Automatic Mixed Precision (AMP) has more benefits in terms of lower GPU memory consumption, faster training execution time, faster convergence towards solutions, and finally, higher accuracy.

关键词： Automatic Mixed Precision Convolutional neural network distributed Data-Parallel Graphics processing Unit Remote Sensing

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：