检索结果-内蒙古大学图书馆

encoder-decoder networks for Analyzing Thermal and Power Delivery networks

ACM TRANSACTIONS ON DESIGN AUTOMATION OF ELECTRONIC SYSTEMS 2023年第1期28卷 3-3页

作者： Chhabria, Vidya A. Ahuja, Vipul Prabhu, Ashwath Patil, Nikhil Jain, Palkesh Sapatnekar, Sachin S. Univ Minnesota 200 Union St SE Minneapolis MN 55455 USA Qualcomm Technol Inc Carina West Tower Marathalli Outer Ring Rd Doddan Bengaluru 560037 Karnataka India

Power delivery network (PDN) analysis and thermal analysis are computationally expensive tasks that are essential for successful integrated circuit (IC) design. Algorithmically, both these analyses have similar computational structure and complexity as they involve the solution to a partial differential equation of the same form. This article converts these analyses into image-to-image and sequence-to-sequence translation tasks, which allows leveraging a class of machine learning models with an encoder-decoder-based generative (EDGe) architecture to address the time-intensive nature of these tasks. For PDN analysis, we propose two networks: (i) IREDGe: a full-chip static and dynamic IR drop predictor and (ii) EMEDGe: electromigration (EM) hotspot classifier based on input power, power grid distribution, and power pad distribution patterns. For thermal analysis, we propose ThermEDGe, a full-chip static and dynamic temperature estimator based on input power distribution patterns for thermal analysis. These networks are transferable across designs synthesized within the same technology and packing solution. The networks predict on-chip IR drop, EM hotspot locations, and temperature in milliseconds with negligibly small errors against commercial tools requiring several hours.

关键词： Machine learning (ML) for electronic design automation (EDA) Thermal analysis Power delivery network (PDN) analysis IR drop electromigration (EM) encoder-decoder networks U-Nets

来源：评论

学校读者我要写书评

暂无评论

Learning From Architectural Redundancy: Enhanced Deep Supervision in Deep Multipath encoder-decoder networks

引用

IEEE TRANSACTIONS ON NEURAL networks AND LEARNING SYSTEMS 2022年第9期33卷 4271-4284页

作者： Luo, Ying Lu, Jinhu Jiang, Xiaolong Zhang, Baochang Beihang Univ Sch Automat Sci & Elect Engn Beijing 100191 Peoples R China Beihang Univ Shenyuan Honors Coll Beijing 100191 Peoples R China Beihang Univ Sch Automat Sci & Elect Engn State Key Lab Software Dev Environm Beijing 100191 Peoples R China Beihang Univ Beijing Adv Innovat Ctr Big Data & Brain Comp Beijing 100191 Peoples R China Alibaba Grp Beijing 100102 Peoples R China Beihang Univ Inst Artificial Intelligence Beijing 100191 Peoples R China

Deep encoder-decoders are the model of choice for pixel-level estimation due to their redundant deep architectures. Yet they still suffer from the vanishing supervision information issue that affects convergence because of their overly deep architectures. In this work, we propose and theoretically derive an enhanced deep supervision (EDS) method which improves on conventional deep supervision (DS) by incorporating variance minimization into the optimization. A new structure variance loss is introduced to build a bridge between deep encoder-decoders and variance minimization, and provides a new way to minimize the variance by forcing different intermediate decoding outputs (paths) to reach an agreement. We also design a focal weighting strategy to effectively combine multiple losses in a scale-balanced way, so that the supervision information is sufficiently enforced throughout the encoder-decoders. To evaluate the proposed method on the pixel-level estimation task, a novel multipath residual encoder is proposed and extensive experiments are conducted on four challenging density estimation and crowd counting benchmarks. The experimental results demonstrate the superiority of our EDS over other paradigms, and improved estimation performance is reported using our deeply supervised encoder-decoder.

关键词： Estimation Task analysis Decoding Optimization Redundancy Computer architecture Semantics Convolutional neural networks (CNN) crowd counting deep supervised learning encoder-decoder networks

来源：评论

学校读者我要写书评

暂无评论

Uncertainty quantification for goal-oriented inverse problems via variational encoder-decoder networks

引用

INVERSE PROBLEMS 2024年第7期40卷 075010-075010页

作者： Afkham, Babak Maboudi Chung, Julianne Chung, Matthias Tech Univ Denmark Dept Appl Math & Comp Sci DTU Compute Lyngby Denmark Emory Univ Dept Math Atlanta GA 30322 USA

In this work, we describe a new approach that uses variational encoder-decoder (VED) networks for efficient uncertainty quantification for goal-oriented inverse problems. Contrary to standard inverse problems, these approaches are goal-oriented in that the goal is to estimate some quantities of interest (QoI) that are functions of the solution of an inverse problem, rather than the solution itself. Moreover, we are interested in computing uncertainty metrics associated with the QoI, thus utilizing a Bayesian approach for inverse problems that incorporates the prediction operator and techniques for exploring the posterior. This may be particularly challenging, especially for nonlinear, possibly unknown, operators and nonstandard prior assumptions. We harness recent advances in machine learning, i.e. VED networks, to describe a data-driven approach to large-scale inverse problems. This enables a real-time uncertainty quantification for the QoI. One of the advantages of our approach is that we avoid the need to solve challenging inversion problems by training a network to approximate the mapping from observations to QoI. Another main benefit is that we enable uncertainty quantification for the QoI by leveraging probability distributions in the latent and target spaces. This allows us to efficiently generate QoI samples and circumvent complicated or even unknown forward models and prediction operators. Numerical results from medical tomography reconstruction and nonlinear hydraulic tomography demonstrate the potential and broad applicability of the approach.

关键词： deep learning regularization encoder-decoder networks uncertainty quantification quantity of interest hyperparameter selection

来源：评论

学校读者我要写书评

暂无评论

Concrete roadway crack segmentation using encoder-decoder networks with range images

引用

AUTOMATION IN CONSTRUCTION 2020年 120卷 103403-103403页

作者： Zhou, Shanglian Song, Wei Univ Alabama Dept Civil Construct & Environm Engn Tuscaloosa AL 35487 USA

Recently, researchers have utilized DCNN for pixel-wise crack classification through semantic segmentation. Nevertheless, some issues in current DCNN-based roadway crack segmentation are yet to be fully addressed. For example, image pre-processing techniques are often required to eliminate the surface variations in range images, which may bring uncertainties due to subjective parameter selection;besides, disturbances from many non-crack patterns such as pavement grooves can deteriorate the crack segmentation performance, which remains a challenge for current DCNN-based methodologies. This paper proposes a methodology based on encoder-decoder networks to achieve pixel-wise crack classification performance on laser-scanned range images, under the disturbance of surface variations and grooved patterns in concrete pavements. The raw range data is directly applied in this methodology without any pre-processing. A comparative study is performed to determine the optimal architecture layout among twelve proposed candidates. Meanwhile, the influence of residual connections on DCNN performance is investigated and demonstrated.

关键词： Deep convolutional neural network encoder-decoder networks Residual connection Range image Crack segmentation Pavement groove

来源：评论

学校读者我要写书评

暂无评论

Timber Tracing with Multimodal encoder-decoder networks 18th

Timber Tracing with Multimodal Encoder-Decoder Networks

引用

18th international Conference on Computer Analysis of Images and Patterns (CAIP)

作者： Zolotarev, Fedor Eerola, Tuomas Lensu, Lasse Kalviainen, Heikki Haario, Heikki Heikkinen, Jere Kauppi, Tomi Lappeenranta Lahti Univ Technol LUT Sch Engn Sci Dept Computat & Proc Engn Machine Vis & Pattern Recognit Lab POB 20 Lappeenranta 53851 Finland Finnos Oy Tukkikatu 5 Lappeenranta 53900 Finland

ISBN: (纸本)9783030298913;9783030298906

Tracking timber in the sawmill environment from the raw material (logs) to the end product (boards) provides various benefits including efficient process control, the optimization of sawing, and the prediction of end-product quality. In practice, the tracking of timber through the sawmilling process requires a methodology for tracing the source of materials after each production step. The tracing is especially difficult through the actual sawing step where a method is needed for identifying from which log each board comes from. In this paper, we propose an automatic method for board identification (board-to-log matching) using the existing sensors in sawmills and multimodal encoder-decoder networks. The method utilizes point clouds from laser scans of log surfaces and grayscale images of boards. First, log surface heightmaps are generated from the point clouds. Then both the heightmaps and board images are converted into "barcode" images using convolutional encoder-decoder networks. Finally, the "barcode" images are utilized to find matching logs for the boards. In the experimental part of the work, different encoder-decoder architectures were evaluated and the effectiveness of the proposed method was demonstrated using challenging data collected from a real sawmill.

关键词： Convolutional neural networks encoder-decoder networks Multimodal translation Machine vision Sawmilling

来源：评论

学校读者我要写书评

暂无评论

A DEEP encoder-decoder networks FOR JOINT DEBLURRING AND SUPER-RESOLUTION

A DEEP ENCODER-DECODER NETWORKS FOR JOINT DEBLURRING AND SUP...

引用

IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

作者： Zhang, Xinyi Wang, Fei Dong, Hang Guo, Yu Xi An Jiao Tong Univ Inst Artificial Intelligence & Robot Xian Shaanxi Peoples R China

ISBN: (纸本)9781538646588

In this paper, we propose an end-to-end convolution neural network (CNN) to restore a clear high-resolution image from a severely blurry image. It's a highly ill-posed problem and brings tremendous challenges to state-of-art deblurring or super-resolution (SR) methods. A straightforward way to solve this problem is to concatenate two types of networks directly. However, experiments show that the concatenation of independent networks increases computation complexity instead of generating satisfying high-resolution images. Consequently, we focus on designing a single deep network to solve the deblurring and SR problems in parallel. Our method, called ED-DSRN, extends the traditional Super-Resolution network by adding a deblurring branch that shares the same feature maps extracted from an encoder-decoder module with the original SR branch. Extensive experiments show that our method produces remarkable deblurred and super-resolved images simultaneously with high efficiency.

关键词： Super-Resolution Blind deblurring Joint tasks encoder-decoder networks Parallel branches

来源：评论

学校读者我要写书评

暂无评论

A Deep encoder-decoder networks for Joint Deblurring and Super-Resolution

A Deep Encoder-Decoder Networks for Joint Deblurring and Sup...

引用

IEEE International Conference on Acoustics, Speech and Signal Processing

作者： Xinyi Zhang Fei Wang Hang Dong Yu Guo Institute of Artificial Intelligence and Robotics Xi'an Jiaotong University China

ISBN: (纸本)9781538646595

关键词： Super-Resolution Blind deblurring Joint tasks encoder-decoder networks Parallel branches

来源：评论

学校读者我要写书评

暂无评论

Creating an AI fashioner through deep learning and computer vision

引用

EVOLVING SYSTEMS 2024年第3期15卷 717-729页

作者： Balim, Caner Ozkan, Kemal Afyon Kocatepe Univ Sandikli Vocat Sch Higher Educ Dept Comp Comp Programming Afyonkarahisar Turkiye Eskisehir Osmangazi Univ Fac Engn & Architecture Dept Comp Engn Eskisehir Turkiye

Fashion is a multibillion-dollar industry that concerns many people both socially and culturally. Thanks to social networks, there is a lot of data about the fashion industry on the internet. This has led researchers to shift their attention to this area, especially recently. This paper proposes an end-to-end framework to build an AI fashioner that can diagnose clothing compatibility and generate recommendations to improve compatibility. First, fashion compatibility reviews are analyzed, and incompatible clothing items are identified for each outfit combination. Next, the items of clothing that make up the outfit combination are separated using Mask R-CNN. Then, the incompatible clothing items were removed from the outfit combination, and the most similar outfit combinations were identified among the compatible clothing items. In addition, an attribute detection network was developed to extract the attributes of compatible outfits with the same category in the detected compatible outfits. Finally, recommendation sentences are generated using the detected attributes, and encoder-decoder models are used to train a deep network that generates recommendations from clothing images. Extensive experiments based on existing datasets demonstrate the effectiveness of the proposed method.

关键词： Outfit recommendation encoder-decoder networks Outfit segmentation Image captioning Image processing

来源：评论

学校读者我要写书评

暂无评论

Plant leaf deep semantic segmentation and a novel benchmark dataset for morning glory plant harvesting

引用

NEUROCOMPUTING 2023年第1期555卷

作者： Su, Jingxuan Anderson, Sean Javed, Mahed Khompatraporn, Charoenchai Udomsakdigool, Apinanthana Mihaylova, Lyudmila Univ Sheffield Dept Automat Control & Syst Engn Sheffield S10 3JD England King Mongkuts Univ Technol Thonburi KMUTT Dept Prod Engn Bangkok Thailand

Computer vision and deep learning have made substantial progress in the areas of agriculture and smart farming, particularly for enhancing crop production using image segmentation techniques for crop yield prediction. Further improvements to crop yield prediction results can be achieved by developing accurate and efficient methods. In response to such demands, this paper proposes a novel convolutional neural network architecture, called densely connected SegNet (D-SegNet) and demonstrates its advantages on plant segmentation using a new morning glory plant dataset, and also on a complimentary publicly available dataset to promote research in this direction. The D-SegNet is evaluated using 10-fold cross validation. It achieves performance better than the state-of-the-art SegNet algorithm. The evaluated precision, recall and F1-score values are 98.20%, 90.64% and 94.26%, respectively, for the morning glory plant dataset. The intersection over union (IoU) value in the image segmentation tasks is 90.56%. A series of experiments on the morning glory plant dataset as well as on the publicly available dataset were conducted. The results show that the proposed method achieves accurate segmentation results and can be useful for assessing the plant weight during harvesting. In summary, this new plant segmentation network, D-SegNet, could form an important component of future cloud-based machine learning systems to predict crop yield from noisy smartphone images taken in the field.

关键词： Semantic image segmentation Yield prediction Precision agriculture Deep learning methods encoder-decoder networks Dense block

来源：评论

学校读者我要写书评

暂无评论

EMED-UNet: An Efficient Multi-encoder-decoder Based UNet for Medical Image Segmentation

引用

IEEE ACCESS 2023年 11卷 95253-95266页

作者： Shah, Kashish D. Patel, Dhaval K. Thaker, Minesh P. Patel, Harsh A. Saikia, Manob Jyoti Ranger, Bryan J. Ahmedabad Univ Sch Engn & Appl Sci Ahmadabad 380009 India Indus Univ Indus Inst Technol & Engn Ahmadabad 382115 India Univ North Florida Dept Elect Engn Jacksonville FL 32224 USA Boston Coll Chestnut Hill Newton MA 02467 USA

Many current and state-of-the-art deep learning models for accurate image segmentation are based on the U-Net architecture, a convolutional neural network designed for biomedical applications. Despite its widespread adoption in the medical imaging community, U-Net has two major limitations. First, due to its deep structure and the large number of filters used, the number of parameters and Floating-Point Operations Per Second (FLOPS) are high. This results in high computational complexity and demands a large memory size, making real-time implementation and deployment of U-Net models challenging. Second, the base U-Net model only uses a single kernel type (i.e., $3\times 3$ ) throughout the network for all convolution operations. Feature extraction using a single spatial extent throughout the network is suitable if the size, location, and shape of the salient regions remain static in all the dataset images, which is not necessarily the case in medical imaging. To address these two limitations, we propose an Efficient Multi-encoder-decoder based UNet (EMED-UNet), a novel architecture for efficient medical image segmentation. We evaluated our network on four medical imaging datasets: Montgomery County, Shenzhen CXR, COVID-19 CT LS, and the BraTS (Brain Tumor Segmentation) dataset. EMED-UNet outperforms U-Net and its variants in terms of accuracy, with around 77% reduction in parameters, a 60% reduction in FLOPS, and a 79.2% reduction in memory usage (all as compared to U-Net). The results demonstrate that EMED-UNet is a lightweight and accurate model for image segmentation that substantially improves upon the U-Net base model and is more feasible to deploy, given its decreased computational cost.

关键词： Deep learning U-Net bottleneck layer semantic segmentation medical image segmentation encoder-decoder networks

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：