检索结果-内蒙古大学图书馆

Design, Automation and Test in Europe Conference and Exhibition

作者： Salma Afifi Ishan Thakkar Sudeep Pasricha Department of Electrical and Computer Engineering Colorado State University Fort Collins USA Department of Electrical and Computer Engineering University of Kentucky Lexington USA

ISBN: (数字)9783982674100

ISBN: (纸本)9798331534646

The rapid proliferation of deep learning has revolutionized computing hardware, driving innovations to improve computationally expensive multiply-accumulate operations in deep neural networks. Among these innovations are integrated silicon-photonic systems that have emerged as energy-efficient platforms capable of achieving light speed computation and communication, positioning optical neural network (ONN) platforms as a transformative technology for accelerating deep learning models such as convolutional neural networks (CNNs). However, the increasing complexity of optical hardware introduces new vulnerabilities, notably the risk of hardware trojan (HT) attacks. Despite the growing interest in ONN platforms, little attention has been given to how HT-induced threats can compromise performance and security. This paper presents an in-depth analysis of the impact of such attacks on the performance of CNN models accelerated by ONN accelerators. Specifically, we show how HTs can compromise microring resonators (MRs) in a state-of-the-art non-coherent ONN accelerator and reduce classification accuracy across CNN models by up to 7.49% to 80.46% by just targeting 10% of MRs. We then propose techniques to enhance ONN accelerator robustness against these attacks and show how the best techniques can effectively recover the accuracy drops.

关键词： Deep learning Technological innovation Analytical models Accuracy optical resonators optical computing optical fiber networks Convolutional neural networks Trojan horses Security

来源：评论

学校读者我要写书评

暂无评论

MLNet: multichannel feature fusion lozenge network for land segmentation

引用

JOURNAL OF APPLIED REMOTE SENSING 2022年第1期16卷

作者： Gao, Jiahong Weng, Liguo Xia, Min Lin, Haifeng Nanjing Univ Informat Sci & Technol Jiangsu Collaborat Innovat Ctr Atmospher Environm Nanjing Peoples R China Nanjing Univ Informat Sci & Technol B DAT Nanjing Peoples R China Nanjing Forestry Univ Coll Informat Sci & Technol Nanjing Peoples R China

The use of remote sensing images for land cover analysis has broad prospects. At present, the resolution of aerial remote sensing images is getting higher and higher, and the span of time and space is getting larger and larger, therefore segmenting target objects enconter great difficulties. Convolutional neural networks are widely used in many image semantic segmentation tasks, but existing models often use simple accumulation of various convolutional layers or the direct stacking of interfeature reuse of up- and downs ampling, the network very heavy. To improve the accuracy of land cover segmentation, we propose a multichannel feature fusion lozenge network. The multichannel feature fusion lozenge network (MLNet) is a three-sided network composed of three branches: one branch uses different levels of feature indexes to sample to maintain the integrity of high-frequency information;one branch focuses on contextual information and strengthens the compatibility of information within and between classes;and the last branch uses feature integration to filter redundant information based on multiresolution segmentation to extract key features. Compared with FCN, UNet, PSP, and other serial single road computing models, the MLNet, which performs feature fusion after three-way parallelism structure, can significantly improve the accuracy with only small increase in complexity. Experimental results show that the average accuracy of 85.30% is obtained on the land cover data set, which is much higher than that of 82.98% of FCN, 81.87% of UNet, 77.52% of SegNet, and 83.09% of EspNet, which proves the effectiveness of the model. (C) 2022 Society of Photo-optical Instrumentation Engineers (SPIE)

关键词： land cover lozenge net three-sided network multiresolution segmentation extract key features

来源：评论

学校读者我要写书评

暂无评论

GHOST: A Graph neural network Accelerator using Silicon Photonics

GHOST: A Graph Neural Network Accelerator using Silicon Phot...

引用

作者： Afifi, Salma Sunny, Febin Shafiee, Amin Nikdast, Mahdi Pasricha, Sudeep Colorado State University Department of Electrical and Computer Engineering 1373 campus delivery Fort CollinsCO80523 United States

Graph neural networks (GNNs) have emerged as a powerful approach for modelling and learning from graph-structured data. Multiple fields have since benefitted enormously from the capabilities of GNNs, such as recommendation systems, social network analysis, drug discovery, and robotics. However, accelerating and efficiently processing GNNs require a unique approach that goes beyond conventional artificial neural network accelerators, due to the substantial computational and memory requirements of GNNs. The slowdown of scaling in CMOS platforms also motivates a search for alternative implementation substrates. In this paper, we present GHOST, the first silicon-photonic hardware accelerator for GNNs. GHOST efficiently alleviates the costs associated with both vertex-centric and edge-centric operations. It implements separately the three main stages involved in running GNNs in the optical domain, allowing it to be used for the inference of various widely used GNN models and architectures, such as graph convolution networks and graph attention networks. Our simulation studies indicate that GHOST exhibits at least 10.2 × better throughput and 3.8 × better energy efficiency when compared to GPU, TPU, CPU and multiple state-of-the-art GNN hardware accelerators. © 2023 Copyright held by the owner/author(s).

关键词： Silicon photonics

来源：评论

学校读者我要写书评

暂无评论

Handwritten Digit Recognition using Fine-Tuned Convolutional neural network Model 3

Handwritten Digit Recognition using Fine-Tuned Convolutional...

引用

3rd International Conference on Smart Technologies, Communication and Robotics, STCR 2023

作者： Srivastav, Somya Guleria, Kalpna Sharma, Shagun Chitkara University Institute of Engineering and Technology Chitkara University Punjab India

ISBN: (纸本)9798350370867

Traditional handwriting recognition algorithms were reliant heavily on existing knowledge and personalized features. It is more difficult to train a computerized optical character recognition (OCR) system using these criteria. Recent advances in the identification of handwriting research have been made possible by using deep learning techniques as their main mechanism. But when the volume of written content quickly increased and incredibly powerful computing resources became available, digit recognition accuracy had to have deteriorated, mandating more applicants' implementation in order to yield accurate and precise results. To develop an effective approach for resolving handwriting detection challenges, this work uses the convolutional neural network (CNN) model which demonstrated the outperforming results in interpreting the structure of handwritten characters as well as phrases. This model is also capable of reducing the operational complexity and cost by implementing this model on different optimizers namely, Adam and SGD. The results have been evaluated at different numbers of epochs which identified that the proposed model outperforms with Adam optimizer by showing the highest accuracy of 98% at epoch 50. This model can be further implemented by varying the learning rate values to achieve improved accuracy and reducing the loss. © 2023 IEEE.

关键词： neural network models

来源：评论

学校读者我要写书评

暂无评论

Short-Term forecast of port cargo throughput based on ARIMA-RBF neural network

Short-Term forecast of port cargo throughput based on ARIMA-...

引用

2021 International Conference on High Performance computing and Communication, HPCCE 2021

作者： Shen, Yongcen Transportation Shanghai Maritime University Shanghai201306 China

ISBN: (数字)9781510652019

ISBN: (纸本)9781510652002

Based on the historical data of Jiaxing Port, this paper uses ARIMA-RBF to predict the monthly cargo throughput of Jiaxing Port, builds an ARIMA-RBF model, and finds the optimal network parameters through training. On this basis, the ARIMA-RBF model is used to predict the cargo throughput of Jiaxing Port, and the actual data analysis is used to compare the ARIMA-RBF neural network, grey system theory and ARIMA prediction results. Finally, the prediction results of different models are compared to prove that the ARIMA-RBF neural network prediction is the optimal prediction model. © COPYRIGHT SPIE. Downloading of the abstract is permitted for personal use only.

关键词： Forecasting

来源：评论

学校读者我要写书评

暂无评论

Image Inpainting With Deep Adaptive Generative models for Multimedia computing 10

Image Inpainting With Deep Adaptive Generative Models for Mu...

引用

Optoelectronic Imaging and Multimedia Technology X 2023

作者： Voronin, V. Gapon, N. Zhdanova, M. Tokareva, O. Semenishchev, E. Center for Cognitive Technology and Machine Vision Moscow State University of Technology «STANKIN» Moscow Russia Don State Technical University Rostov-on-Don Russia

ISBN: (纸本)9781510667839

Image inpainting is a challenging task where large missing regions must be filled based on the available data. It is an interesting new research topic in multimedia computing and image processing. This paper proposes a novel deep learning reconstruction method using an adaptive generative adversarial neural network (aGAN). The key feature of the proposed network is the dynamic estimation of the trend of the loss function. This estimate finds the optimal balance between adversarial and pixel-by-pixel losses. We introduced the coefficient of the approximating curve, which allows us to estimate the trend dynamics of the loss function. Depending on the value of this curve, we adjust the weight of the pixel-by-pixel loss to make the restored area sharper. The proposed method allows a higher sharpness of the reconstructed area and a more stable learning process. Experiments on the dataset show that our method successfully predicts multimedia information in missing regions and significantly outperforms the state-of-the-art methods. © 2023 SPIE. All rights reserved.

关键词： Pixels

来源：评论

学校读者我要写书评

暂无评论

Early Prediction and Analysis of Pancreatitis Laboratory Data Based on neural network 2

Early Prediction and Analysis of Pancreatitis Laboratory Dat...

引用

2nd International Conference on Biomedical and Intelligent Systems, IC-BIS 2023

作者： Tong, Huifeng Zhou, Hua College of Big Data and Intelligent Engineering Southwest Forestry University Kunming650224 China

ISBN: (纸本)9781510666733

Pancreatitis is a serious pancreatic disease. Due to its inconspicuous early symptoms, it is easy to be ignored or misdiagnosed, causing many patients to miss the best time for treatment, and even life-threatening. According to statistics, the prevalence of pancreatitis is increasing year by year, and it has become one of the diseases that endanger human health. Therefore, early warning has become an important link in the treatment of pancreatitis, and laboratory test data is a reliable predictor. In this study, an early warning system for laboratory data was constructed based on neural networks to predict the occurrence of pancreatitis. The neural network model uses a 3-layer feed-forward neural network and uses the backpropagation algorithm for training. The experimental data comes from the laboratory test results of 90 patients with pancreatitis. Predictive indicators included 24 blood cell analysis data. Through the model verification experiment, this study achieved good prediction results, proved the feasibility and prediction accuracy of the model, and provided ideas and basis for the practice of computer-aided diagnosis of late pancreatitis. In view of the limited sample data, the experiment only The two-category research is limited. In order to better expand, the sample data will be expanded later to cover common diagnostic types and improve the generalization ability of the model. © 2023 SPIE.

关键词： neural network models

来源：评论

学校读者我要写书评

暂无评论

An optical Satellite Controller Based on Diffractive Deep neural network 2nd

An Optical Satellite Controller Based on Diffractive Deep Ne...

引用

2nd CAAI International Conference on Artificial Intelligence (CICAI)

作者： Liu, Shaohua Dou, Hongkun Li, Hongjue Deng, Yue Beihang Univ Sch Astronaut Beijing Peoples R China

ISBN: (纸本)9783031204999;9783031205002

Formulated as an optimal control problem, space relative trajectory planning is crucial for on-orbit servicing spacecraft on various missions. While a variety of deep-neural-network (DNN) methods have been proposed to solve the problem, they are energy-consuming and computationally consuming, which limits their on-board deployments. In this work, we proposed a kind of diffractive-deep-neural-network-based optical satellite controller, and transformed the solution of the optimal control problem into a light-field-based fine-grained regression task. Firstly, an electro-optic conversion module was designed to convert numerical relative state variables from electronic signals into light field as the input of a diffractive modulation (DM) module, the diffractive masks of which could be trained to implement complex light field transformation. We used another optic-electro conversion module to convert the light field at the output plane of DM module into electronic signals. Then, we trained the DM module to make the decoded electrical signals consistent with the desired optimal control commands. Therefore, when the light carrying input information and propagating through the well-trained diffractive masks, the DM module could perform diffract-based solution of optimal control problem. The simulation results substantiate the feasibility and effectiveness of our OC-Nets, which can achieve comparable performance to the latest classic DNN methods, except for a few acceptable errors. Different from classic models with much too energy consumption, once fabricated physically, the device of our optical controller can provide optimal control commands at the speed of light, with fairly little computational and energy consumption, and enable the on-board deployment on spacecraft.

关键词： optical computing Diffractive deep neural network On-orbit service Optimal control

来源：评论

学校读者我要写书评

暂无评论

Trajectory-text retrieval model based on graph neural network 3

Trajectory-text retrieval model based on graph neural networ...

引用

3rd International Conference on Computer Vision and Data Mining, ICCVDM 2022

作者： Tang, Hanchao Wang, Licai Luo, Qibin North China Institute of Computing Technology Haidian District Beijing100089 China

ISBN: (纸本)9781510661363

Cross-modal retrieval has been widely used in the Vision-Language field and has achieved many results, but there is a lack of research in the trajectory-text field. At the same time, the current popular cross-modal retrieval models not only lack fine-grained semantic alignment between different modalities, but also ignore the influence of the grammatical structure of the text on the retrieval effect. To solve the above problems, this paper proposes a dual-stream trajectory text retrieval model combined with graph neural network, combining local and global two cross-modal interaction methods: (1) Local alignment, encoding trajectory points and words respectively after passing through the masking module. Semantic alignment. (2) Global alignment, introducing momentum contrastive learning to achieve trajectory and text retrieval learning. Experimental results show that this hierarchical matching method not only retains the efficient performance of the dual-stream model, but also has higher accuracy than other cross-modal retrieval models, and its R@1 value on the dataset is improved by 3.2%-4.7%. © COPYRIGHT SPIE. Downloading of the abstract is permitted for personal use only.

关键词： Trajectories

来源：评论

学校读者我要写书评

暂无评论

A Hybrid-Attention-LSTM-Based Deep Convolutional neural network to Extract Modal Frequencies from Limited Data Using Transfer Learning 40th

A Hybrid-Attention-LSTM-Based Deep Convolutional Neural Netw...

引用

40th Conference and Exposition on Structural Dynamics (IMAC)

作者： Dizaji, Mehrdad Shafiei Mao, Zhu Univ Massachusetts Lowell Dept Mech Engn Lowell MA 01854 USA Worcester Polytechn Inst Dept Mech & Mat Engn Worcester MA USA

ISBN: (纸本)9783031040986;9783031040979

Current computer video-based vibration modal analysis approaches typically decompose video frames into representations and then adjust them that allow to magnify motions to extract motion representations for vibration modal analysis. Their decomposition usually relies upon handcrafted designed kernels, such as the complex steerable kernels, which typically may not be optimally designed for the extraction of subtle motions specially in higher frequency domains. In this paper, optimal decomposition kernel is learned and designed directly from baseline dataset images using deep convolutional neural network (CNN) models. Each subpixel of an image obtained from a digital camera is included when computing the spatiotemporal information, which serves similar to an individual motion sensor to acquire the modal frequencies of a vibrating structure. A hybrid-attention-LSTM-based deep convolutional neural network architecture is developed to take advantage of attention and LSTM blocks to discover subtle motions from a specific source to visualize high resolution of dynamic properties of the structures in the existence of high amounts of noise. The idea of transfer learning is utilized to transfer the knowledge previously learned to new limited dataset. Transfer learning is used to take advantage of limited existing dataset to avoid underfitting in the training of the network, considering the current publicly available modal frequency datasets are insufficient to train a generalized network. The proposed deep learning architecture is designed in such a way that has capability of transferring the trained model from baseline dataset on a simple structure to a complicated structure using transfer learning perspective. After training, the model takes the video of a vibrating structure as input and outputs the fundamental modal frequencies. By showing reliable empirical results, the proposed model is autonomous, efficient, and accurate.

关键词： Convolutional neural networks Deep learning Transfer learning Modal analysis

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：