检索结果-内蒙古大学图书馆

17th International Conference on Intelligent Robotics and Applications

作者： Yang, Feng Gao, Tongyang Zheng, Litao Liao, Pan Northwestern Polytech Univ Sch Automat Xian Peoples R China Shanghai Jiao Tong Univ Dept Automat Shanghai Peoples R China

ISBN: (纸本)9789819607853;9789819607860

Traditional model-based distributed maneuvering target tracking methods often require prior knowledge of both sensor and target models, which can lead to significant degradation in tracking accuracy when faced with model mismatches. To address this problem, this paper presents a novel model-based deep learning distributed maneuvering target tracking method over a homogeneous sensor network. Specifically, two deep neural networks are designed to obtain noise-eliminated measurements and timely estimate the motion model of the target. Further, hybrid consensus on measurements and consensus on information method is utilized to fuse local filtering results. Finally, the simulation experiment demonstrates the effectiveness of the proposed method.

关键词： Distributed Sensor Network model-based deep learning Maneuvering Target Tracking

来源：评论

学校读者我要写书评

暂无评论

Efficient model-based deep learning via Network Pruning and Fine-Tuning

引用

JOURNAL OF MATHEMATICAL IMAGING AND VISION 2025年第2期67卷 1-14页

作者： Park, Chicago Y. Gan, Weijie Zou, Zihao Hu, Yuyang Sun, Zhixin Kamilov, Ulugbek S. Washington Univ St Louis Dept Comp Sci & Engn 1 Brookings Dr St Louis MO 63130 USA Washington Univ St Louis Dept Elect & Syst Engn 1 Brookings Dr St Louis MO 63130 USA

model-based deep learning (MBDL) is a powerful methodology for designing deep models to solve imaging inverse problems. MBDL networks can be seen as iterative algorithms that estimate the desired image using a physical measurement model and a learned image prior specified using convolutional neural networks (CNNs). The iterative nature of MBDL networks increases the test-time computational complexity, which limits their applicability in certain large-scale applications. Here we make two contributions to address this issue: First, we show how structured pruning can be adopted to reduce the number of parameters in MBDL networks. Second, we present three methods to fine-tune the pruned MBDL networks to mitigate potential performance loss. Each fine-tuning strategy has a unique benefit that depends on the presence of a pre-trained model and a high-quality ground truth. We show that our pruning and fine-tuning approach can accelerate image reconstruction using popular deep equilibrium learning (DEQ) and deep unfolding (DU) methods by 50% and 32%, respectively, with nearly no performance loss. This work thus offers a step forward for solving inverse problems by showing the potential of pruning to improve the scalability of MBDL. Code is available at https://***/wustl-cig/MBDL_Pruning.

关键词： Computational imaging Inverse problem model-based deep learning Network pruning

来源：评论

学校读者我要写书评

暂无评论

model-based deep learning Algorithm for Detection and Classification at High Event Rates

引用

IEEE TRANSACTIONS ON NUCLEAR SCIENCE 2024年第5期71卷 970-980页

作者： Morad, Itai Ghelman, Max Ginzburg, Dimitry Osovizky, Alon Shlezinger, Nir Ben Gurion Univ Negev Sch ECE IL-84105 Beer Sheva Israel Israel Atom Energy Commiss IL-61070 Tel Aviv Israel Nucl Res Ctr Elect & Control Labs IL-84190 Negev Israel Rotem Ind Ltd Hlth Phys Instrumentat Dept IL-85339 Lehavim Israel Nucl Res Ctr Elect & Control Labs Beer Sheva Israel

Pulse shape discrimination (PSD) is required for many radioactive particle monitoring applications. Classical PSD methods commonly struggle at high event rates in the presence of pile up, and are therefore utilized for low event rates. We present a PSD algorithm that combines classic approaches with deep learning techniques. The algorithm provides both detection and classification of the pulses at high event rates. While PSD algorithms for high event rates are often limited to two piled-up pulses, our algorithm is designed and tested for detection and classification under severe pile-up conditions, where three or more pulses are piled up. The algorithm was tested on both experimental data from a Cs(2)LiYCl6:Ce (CLYC) detector and on synthetic data. The algorithm's detection and discrimination performance is compared to current state-of-the-art methods. The algorithm's performance is characterized under varying event rates, signal-to-noise ratio (SNR) conditions, and neutron-to-gamma event rate ratios.

关键词： Cs(2)LiYCl6:Ce (CLYC) high rate model-based deep learning pulse pile up pulse shape discrimination (PSD)

来源：评论

学校读者我要写书评

暂无评论

model-based deep learning for One-Bit Compressive Sensing

引用

IEEE TRANSACTIONS ON SIGNAL PROCESSING 2020年 68卷 5292-5307页

作者： Khobahi, Shahin Soltanalian, Mojtaba Univ Illinois Dept Elect & Comp Engn Chicago IL 60607 USA

In this work, we consider the problem of one-bit deep compressive sensing from both a system design and a signal recovery perspective. In particular, we develop hybrid model-based deep learning architectures based on the deep unfolding methodology. We further interpret the overall data-acquisition and signal recovery modules as an auto-encoder structure allowing for learning task-specific sensing matrix, quantization thresholds, as well as the latent-parameters of iterative first-order optimization algorithms specifically designed for the problem of one-bit sparse signal recovery. The proposed model-based deep architectures have the ability to adaptively learn the proper quantization thresholds, paving the way for amplitude recovery in one-bit compressive sensing. We further show that the proposed methodology implicitly learns task-specific sensing matrices with very low coherence, which is highly desirable in a compressive sensing setting. Due to the model-based nature of the proposed deep architecture, it enjoys from the interpretability and versatility of model-based techniques as well as benefiting from the expressive power of data-driven methods. Specifically, owing to its model-based nature, it has far fewer parameters and requires far less samples for training as compared to black-box machine learning models. Our results demonstrate a significant improvement compared to state-of-the-art algorithms.

关键词： Autoencoders compressive sensing deep neural networks deep unfolding low-resolution signal processing model-based deep learning one-bit quantization

来源：评论

学校读者我要写书评

暂无评论

model-based deep learning for additive manufacturing: New frontiers and applications

引用

MANUFACTURING LETTERS 2021年 29卷 94-98页

作者： Ghungrad, Suyog Gould, Benjamin Soltanalian, Mojtaba Wolff, Sarah Jeannette Haghighi, Azadeh Univ Illinois Dept Mech & Ind Engn Chicago IL 60607 USA Argonne Natl Lab Appl Mat Div 9700 S Cass Ave Lemont IL 60439 USA Univ Illinois Dept Elect & Comp Engn Chicago IL 60607 USA Texas A&M Univ Ind & Syst Engn College Stn TX 77843 USA

Artificial intelligence has created disruptive possibilities in additive manufacturing towards smarter design, process control, and quality assurance. Nonetheless, the scarcity of data in additive manufacturing significantly limits the wide adoption of artificial intelligent techniques. In this work, we propose the deployment of a novel artificial intelligent structure called model-based deep learning in the context of additive manufacturing which can address scenarios with scarce data but an available underlying iterative mathematical/inference model. Several immediate applications of this technique in the additive manufacturing research as well as a proof of concept on temperature profile prediction in metal AM process are presented. (C) 2021 Society of Manufacturing Engineers (SME). Published by Elsevier Ltd. All rights reserved.

关键词： Additive manufacturing model-based deep learning Artificial intelligence Data scarcity Process and material variability

来源：评论

学校读者我要写书评

暂无评论

model-based deep learning for Reconstruction of Joint k-q Under-sampled High Resolution Diffusion MRI 17

Model-Based Deep Learning for Reconstruction of Joint k-q Un...

引用

IEEE 17th International Symposium on Biomedical Imaging (ISBI)

作者： Mani, Merry P. Aggarwal, Hemant K. Ghosh, Sanjay Jacob, Mathews Univ Iowa Iowa City IA 52242 USA

ISBN: (纸本)9781538693308

We propose a model-based deep learning architecture for the reconstruction of highly accelerated diffusion magnetic resonance imaging (MRI) that enables high resolution imaging. The proposed reconstruction jointly recovers all the diffusion weighted images in a single step from a joint k-q under-sampled acquisition in a parallel MRI setting. We propose the novel use of a pre-trained denoiser as a regularizer in a model-based reconstruction for the recovery of highly under-sampled data. Specifically, we designed the denoiser based on a general diffusion MRI tissue microstructure model for multi-compartmental modeling. By using a wide range of biologically plausible parameter values for the multi-compartmental microstructure model, we simulated diffusion signal that spans the entire microstructure parameter space. A neural network was trained in an unsupervised manner using an autoencoder to learn the diffusion MRI signal subspace. We employed the autoencoder in a model-based reconstruction and show that the autoencoder provides a strong denoising prior to recover the q-space signal. We show reconstruction results on a simulated brain dataset that shows high acceleration capabilities of the proposed method.

关键词： K-q space deep learning diffusion MRI autoencoder model-based deep learning

来源：评论

学校读者我要写书评

暂无评论

model-based deep learning: KEY APPROACHES AND DESIGN GUIDELINES

MODEL-BASED DEEP LEARNING: KEY APPROACHES AND DESIGN GUIDELI...

引用

IEEE Data Science and learning Workshop (DSLW)

作者： Shlezinger, Nir Whang, Jay Eldar, Yonina C. Dimakis, Alexandros G. Ben Gurion Univ Negev Sch ECE Beer Sheva Israel Univ Texas Austin Dept CS Austin TX 78712 USA Weizmann Inst Sci Fac Math & CS Rehovot Israel Univ Texas Austin Dept ECE Austin TX 78712 USA

ISBN: (纸本)9781665428255

Signal processing, communications, and control have traditionally relied on classical statistical modeling techniques. Such model-based methods tend to be sensitive to inaccuracies and may lead to poor performance when real systems display complex or dynamic behavior. On the other hand, purely data-driven approaches are becoming increasingly popular. deep neural networks (DNNs) employ a highly flexible function class to learn mappings from data, and demonstrate excellent performance. However, DNNs typically require massive amounts of data and immense computational resources, limiting their applicability for some signal processing scenarios. We consider hybrid techniques that combine principled mathematical models with data-driven systems to benefit from the advantages of both approaches. Such model-based deep learning methods exploit both partial domain knowledge, via mathematical structures designed for specific problems, as well as learning from limited data. Here, we survey leading approaches for studying and designing model-based deep learning systems, along with concrete design guidelines and signal processing oriented examples.

关键词： model-based deep learning

来源：评论

学校读者我要写书评

暂无评论

model-based deep learning for Computational Imaging

Model-Based Deep Learning for Computational Imaging

引用

作者： Xu, Xiaojian Washington University in St. Louis

学位级别：Ph.D., Doctor of Philosophy

This dissertation addresses model-based deep learning for computational imaging. The motivation of our work is driven by the increasing interests in the combination of imaging model, which provides data-consistency guarantees to the observed measurements, and deep learning, which provides advanced prior modeling driven by data. Following this idea, we develop multiple algorithms by integrating the classical model-based optimization and modern deep learning to enable efficient and reliable imaging. We demonstrate the performance of our algorithms by validating their performance on various imaging applications and providing rigorous theoretical analysis. The dissertation evaluates and extends three general frameworks, plug-and-play priors (PnP), regularized by denoising (RED) and deep unfolding (DU), all of which integrate model-based optimization and deep learning. PnP and RED adopt deep-learned denoisers as image priors inside model-based iterative algorithms, while DU interprets the iterations of a model-based algorithm as layers of a deep neural network and trains it end-to-end in a supervised fashion. We contribute to these research areas by 1) providing the statistical interpretation of the PnP algorithms through the analysis of the priors implicitly represented by denoisers; 2) proposing an incremental variant of the widely-used PnP-ADMM algorithm to handle problems involving large-scale measurements; 3) extending the family of PnP algorithms to the non-Euclidean setting based on the general Bregman distance; and 4) developing an end-to-end model-based learning framework for the estimation of quantitative maps from under-sampled, noisy and motion-corrupted MRI data.

关键词： Computational imaging deep unfolding model-based deep learning Optimization Plug-and-play priors Regularized by denoising

来源：评论

学校读者我要写书评

暂无评论

Fast Calculation of Probabilistic Power Flow: A model-based deep learning Approach

引用

IEEE TRANSACTIONS ON SMART GRID 2020年第3期11卷 2235-2244页

作者： Yang, Yan Yang, Zhifang Yu, Juan Zhang, Baosen Zhang, Youqiang Yu, Hongxin Chongqing Univ Sch Elect Engn Chongqing 400044 Peoples R China Univ Washington Dept Elect & Comp Engn Seattle WA 98195 USA State Grid Chongqing Elect Power Co Elect Power Res Inst Power Grid Technol Ctr Chongqing 401123 Peoples R China

Probabilistic power flow (PPF) plays a critical role in power system analysis. However, the high computational burden makes it challenging for the practical implementation of PPF. This paper proposes a model-based deep learning approach to overcome the computational challenge. A deep neural network (DNN) is used to approximate the power flow calculation and is trained according to the physical power flow equations to improve its learning ability. The training process consists of several steps: 1) the branch flows are added into the objective function of the DNN as a penalty term, which improves the approximation accuracy of the DNN;2) the gradients used in the back propagation process are simplified according to the physical characteristics of the transmission grid, which accelerates the training speed while maintaining effective guidance of the physical model;and 3) an improved initialization method for the DNN parameters is proposed to improve the convergence speed. The simulation results demonstrate the accuracy and efficiency of the proposed method in standard IEEE and utility benchmark systems.

关键词： Load flow Training Mathematical model Uncertainty Neural networks Standards Probabilistic logic model-based deep learning initialization method deep neural network probabilistic power flow

来源：评论

学校读者我要写书评

暂无评论

SGD-Net: Efficient model-based deep learning With Theoretical Guarantees

引用

IEEE TRANSACTIONS ON COMPUTATIONAL IMAGING 2021年 7卷 598-610页

作者： Liu, Jiaming Sun, Yu Gan, Weijie Xu, Xiaojian Wohlberg, Brendt Kamilov, Ulugbek S. Washington Univ Dept Elect & Syst Engn St Louis MO 63130 USA Washington Univ Dept Comp Sci & Engn St Louis MO 63130 USA Los Alamos Natl Lab Div Theoret Los Alamos NM 87545 USA

deep unfolding networks have recently gained popularity for solving imaging inverse problems. However, the computational and memory complexity of data-consistency layers within traditional deep unfolding networks scales with the number of measurements, limiting their applicability to large-scale imaging inverse problems. We propose SGD-Net as a new methodology for improving the efficiency of deep unfolding through stochastic approximations of the data-consistency layers. Our theoretical analysis shows that SGD-Net can be trained to approximate batch deep unfolding networks to an arbitrary precision. Our simulations on intensity diffraction tomography and sparse-view computed tomography show that SGD-Net can match the performance of the traditional batch network at a fraction of training and testing complexity.

关键词： model-based deep learning regularization by denoising plug-and-play priors stochastic optimization

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：