检索结果-内蒙古大学图书馆

13th ieee data driven control and learning systems conference, DDCLS 2024

作者： Sun, Mengjing Zhu, Haijiang College of Information & Technology Beijing University of Chemical Technology Beijing100029 China

ISBN: (纸本)9798350361674

With the rapid development of computers and the internet, digital image forgery detection has become one of the important research hot topics in the field of computer vision. In this article, we propose a dual stream network to detect image tampering and locate forged areas. Our network includes the ringed residual U-Net, RGB stream and DCT stream, which is named as a hybrid feature ringed residual U-Net (HFRRU-Net) because of combining frequency domain features with RGB features. Compared to the existing works, HFRRU-Net is an end-to-end training approach without image preprocessing or post-processing, and this network may speed up the training and prediction process. In the experments, the HFRRU-Net is evaluated on five public image forgery datasets, CASIA v1.0, Columbia, In-The- Wild, Nist2016, and Realistic. The results indicate that the F1 scores of the HFRRU-Net outperforms other methods in four datasets. And visual comparisons further demonstrate the proposed method. © 2024 ieee.

关键词： data streams

来源：评论

学校读者我要写书评

暂无评论

C²: Co-design of Robots via Concurrent-Network Coupling Online and Offline Reinforcement learning

C<SUP>2</SUP>: Co-design of Robots via Concurrent-Network Co...

引用

ieee/RSJ International conference on Intelligent Robots and systems (IROS)

作者： Chen, Ci Xiang, Pingyu Lu, Haojian Wang, Yue Xiong, Rong Zhejiang Univ State Key Lab Ind Control & Technol Hangzhou 310027 Peoples R China

ISBN: (纸本)9781665491907

With the increasing computing power, using data-driven approaches to co-design a robot's morphology and controller has become a promising way. However, most existing data-driven methods require training the controller for each morphology to calculate fitness, which is time-consuming. In contrast, the dual-network framework utilizes data collected by individual networks under a specific morphology to train a population network that provides a surrogate function for morphology optimization. This approach replaces the traditional evaluation of a diverse set of candidates, thereby speeding up the training. Despite considerable results, the online training of both networks impedes their performance. To address this issue, we propose a concurrent network framework that combines online and offline reinforcement learning (RL) methods. By leveraging the behavior cloning term in a flexible manner, we achieve an effective combination of both networks. We conducted multiple sets of comparative experiments in the simulator and found that the proposed method effectively addresses issues present in the dual-network framework, leading to overall algorithmic performance improvement. Furthermore, we validated the algorithm on a real robot, demonstrating its feasibility in a practical application.

关键词： Reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Identification of linear data-driven models for large-scale power systems 10

Identification of linear data-driven models for large-scale ...

引用

10th International conference on control, Decision and Information Technologies (CoDIT)

作者： Zuccaro, Elia di Capaci, Riccardo Bacci Pannocchia, Gabriele Panosyan, Ara Univ Pisa Dept Informat Engn Pisa Italy Univ Pisa Dept Civil & Ind Engn Pisa Italy Siemens Energy Grid Consulting Erlangen Germany

ISBN: (纸本)9798350373981;9798350373974

The linearization of the full model of electrical power systems is of great significance for the adoption of linear analysis techniques to examine the system's dynamic characteristics, as well as for the design and tuning of practical controllers. Typically, the state-space model of the power system is first obtained from the time domain model. Linear analysis and controller tuning are then performed utilizing the linear state-space model. This approach however often has several practical limitations, such as the unavailability of a time domain model, when only simulation or measurement data is available, or the lack of linearization capability in the software tool in which the time domain model is available. Moreover, the linearization of the time domain models of large-scale power systems results in very high-dimension state-space models, which greatly complicates further analysis. To this aim, in this paper, suitable linear data-driven models of reduced order are identified for power systems to retain the most relevant modes of oscillations of the original system. A commercial rigorous software is used for the data generation and a well-established Python toolbox is used for the model identification: different models and techniques are applied and then compared in terms of accuracy and simplicity.

关键词： Process modeling power systems modal identification system identification

来源：评论

学校读者我要写书评

暂无评论

A Dynamic Robust Predictive Model for Key Variable Estimation of Hydrogen Production Process 13

A Dynamic Robust Predictive Model for Key Variable Estimatio...

引用

13th ieee data driven control and learning systems conference, DDCLS 2024

作者： Yao, Yating Shao, Weiming Department of Chemical Equipment and Control Engineering Qingdao266580 China

ISBN: (纸本)9798350361674

Hydrogen is an ideal energy storage medium obtained mainly by the natural gas steam reforming hydrogen production process. For safe and efficient production of hydrogen, the concentrations of the gases CH4, CO, CO2, and H2 must be measured accurately in real-time. However, traditional methods of measuring these gas concentrations are time-consuming and expensive, and the existing predictive models have not taken the complicated characteristics of the hydrogen production process into consideration. To this end, we develop a predictive model for key gas concentration estimation named 'dynamic robust variational Bayesian principal component analysis (DRVBPCA)'. The DRVBPCA considers complex features of the hydrogen production process, including process dynamics, transportation delays, and outliers. Concretely, the dynamics and the outliers are modeled by the finite impulse response paradigm and Student's-t distribution, respectively, and the transportation delays are automatically determined using the differential evolution algorithm. Finally, the performance of the DRVBPCA is evaluated by a real-life natural gas steam reforming hydrogen production process and compared with other state-of-the-art models. © 2024 ieee.

关键词： Hydrogen storage

来源：评论

学校读者我要写书评

暂无评论

High-speed Interpretable Representation learning of Melt Pool Signatures in Metal Additive Manufacturing 20

High-speed Interpretable Representation Learning of Melt Poo...

引用

ieee 20th International conference on Automation Science and Engineering (CASE)

作者： Vandecasteele, Mathieu Verhees, Dries Philips, Wilfried Bey-Temsamani, Abdellatif Booth, Brian G. Ghent Univ Imec TELIN IPI B-9000 Ghent Belgium Flanders MAKE B-3920 Lommel Belgium

ISBN: (纸本)9798350358513;9798350358520

Additive Manufacturing (AM), i.e. 3D printing, of metal parts is growing in popularity, but part quality can still be a limiting factor to that growth. Camera-based monitoring systems improve quality by detecting defects on-the-fly, but it often relies on incomplete handcrafted video features, or hard to interpret features derived by data-driven techniques. In this work, we propose a method based on variational autoencoders (VAEs) that produces highly informative and interpretable features for in-situ AM monitoring. Unlike handcrafted features, our technique is data-driven so that it captures all details. Unlike other data-driven methods, our technique produces features that are highly interpretable and correlate to the involved physics in the printing process. To test our technique, an object is printed with deliberate non-nominal layers and recorded at high speed with a monochrome optical camera. The video frames are then used to train a VAE as feature extractor. The VAE-computed features are then input into a classification algorithm to detect print deviations. It is shown that our proposed features outperform state-of-the-art deep learning methods, and handcrafted features, with up to 2.16% improvement, all while preserving feature interpretability, the completeness of data-driven feature extraction, and high computation speed (> 5 kHz).

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

WF-Velo: A New Method for RNA Velocity Analysis Based on Single-Cell data 13

WF-Velo: A New Method for RNA Velocity Analysis Based on Sin...

引用

13th ieee data driven control and learning systems conference, DDCLS 2024

作者： Li, Cheng Cao, Wenjie Zhang, Zihao Tang, Wenbo Shi, Yazhou Zhang, Ben-Gong School of Mathematical & Physical Science Wuhan Textile University Wuhan430200 China Sun Yat-Sen University Guangzhou510275 China

ISBN: (纸本)9798350361674

The direction of cell differentiation is very important to study the process of cell differentiation. There was a strong hypothesis that the direction was fixed in previous studies until the RNA velocity method appearance. It is a powerful tool to analyze cell differentiation. However, there is still some problem of it due to the low expression of unspliced mRNA readings and high randomness. This paper intends to propose a new method called WF-velo which means the whole feature (WF) are used during the process of RNA velocity estimations, while the part feature was used by the traditional method, to overcome the problem of the existed methods. By using the characteristics of the gene in the whole transcription process, the RNA velocity of cells can be estimated more realistically, and the cell differentiation process is more in line with biological phenomena compared with the cellDancer, scvelo, Deepvelo and other RNA velocity estimation algorithms. The pancreatic endocrine development dataset and the mouse hippocampus development dataset are tested by WF-velo. The cell differentiation trajectory shows that it is a good method to estimate the RNA velocity. © 2024 ieee.

关键词： Transcription

来源：评论

学校读者我要写书评

暂无评论

Physics-Informed Bayesian Deep learning for Traffic State Estimation and Uncertainty Quantification 27

Physics-Informed Bayesian Deep Learning for Traffic State Es...

引用

27th ieee International conference on Intelligent Transportation systems, ITSC 2024

作者： Huang, Archie J. Filipovska, Monika University of Connecticut Department of Civil and Environmental Engineering StorrsCT United States

ISBN: (纸本)9798331505929

Traffic state estimation (TSE) is a crucial component for efficient traffic control and management. In the literature, given limited and potentially noisy traffic observations, a variety of model-driven and data-driven TSE approaches are developed to reconstruct traffic state variables, such as vehicle density and velocity. In practice, the confidence levels of the reconstruction output also carry valuable insights for transportation practitioners and engineers to develop traffic control measures. In this work, we propose the adoption of the physics-informed Bayesian deep learning (PIBDL) neural network for traffic state estimation and uncertainty quantification (TSE-UQ). Equipped with the knowledge of flow conservation laws, the physics-informed neural network has the advantage of accurately estimating the traffic states, and the component of Bayesian inference in PIBDL enables UQ of the TSE output. We demonstrate the effectiveness of the proposed approach by designing a case study with a synthetic vehicle density dataset, and comparing the PIBDL performance to a baseline deep learning (DL) neural network in TSE-UQ. The results from the case study demonstrate the capability of PIBDL in TSE-UQ applications. © 2024 ieee.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

Contextual Approaches to data-driven Fault Detection in Solar Photovoltaic System

Contextual Approaches to Data-Driven Fault Detection in Sola...

引用

ieee conference on Evolving and Adaptive Intelligent systems (ieee EAIS)

作者： Baruah, Diganta Roy, Ritocheta Ahmed, Raunak Subbiah, Senthilmurugan Chouhan, Sonali Angappan, Kumaresan Indian Inst Technol Guwahati Sch Energy Sci & Engn Gauhati India Sikkim Manipal Inst Technol Dept IT Majitar 737136 Sikkim India Vellore Inst Technol Sch Comp Sci & Engn Vellore Tamil Nadu India SRM Inst Sci & Technol Sch Comp Chennai Tamil Nadu India Indian Inst Technol Guwahati Dept Elect & Elect Engn Gauhati India

ISBN: (纸本)9798350366235;9798350366242

Renewable energy offers a sustainable solution to climate change and environmental degradation by reducing green-house gas emissions and pollution. Solar energy, in particular, is a promising source due to its global availability, environmentally friendly operation, and declining costs. However, faults in solar photovoltaic (PV) systems can diminish their efficiency and pose safety risks. Detecting and diagnosing these faults is crucial for optimizing performance, minimizing downtime, and ensuring safety in solar PV installations. Contextual data, widely used in fields like NLP, computer vision, and recommender systems, are not as prevalent in the domain of solar PV systems. This paper focuses on leveraging contextual information to enhance fault detection in PV systems using ML-based models. The objective is to explore methods for integrating contextual features into machine learning models for AC power prediction and fault detection in solar PV systems. Three strategies for handling contextual data are compared. The results showed that for AC power prediction, contextual expansion and contextual normalisation performed better than contextual model selection strategy.

关键词： solar photovoltaic system power prediction fault detection data-driven model context

来源：评论

学校读者我要写书评

暂无评论

Safe Reinforcement learning for Risk-Aware UAV-Assisted Mobile Edge Computing Networks 13

Safe Reinforcement Learning for Risk-Aware UAV-Assisted Mobi...

引用

13th ieee data driven control and learning systems conference, DDCLS 2024

作者： Shi, Xiangpei Wang, Yuanda Dong, Lu School of Cyber Science and Engineering Southeast University Nanjing210096 China School of Automation Southeast University Nanjing210096 China

ISBN: (纸本)9798350361674

Mobile Edge Computing (MEC) empowers mobile users (MUs) to execute tasks with high computational demands by harnessing the computational capabilities of the network edge. Concurrently, the unmanned aerial vehicle (UAV) enhances wireless systems with adaptable deployment. Utilizing UAV s equipped with computing resources as edge nodes for ground MUs presents a highly promising paradigm. However, the presence of risk areas within the flight range poses significant challenges to flight safety during task execution. This study investigates a safe reinforcement learning (safe RL) approach specifically designed for UAV-assisted MEC networks with risk awareness. Equipped with computing resources, UAV offers offloading services to MU s while prioritizing flight safety and avoiding prolonged navigation within risk areas. Our goal is to collaboratively optimize user scheduling, UAV flight paths, and resource allocation, aiming to minimize processing latency and reduce UAV entry into risk areas whenever possible. Addressing the dynamic changes in complex environments, this paper integrates reinforcement learning algorithms with the Lagrangian method, introducing a secure computing offloading strategy based on SAC-Lagrangian. Comprehensive simulation results validate the effectiveness of the proposed algorithm in terms of latency management and risk avoidance. © 2024 ieee.

关键词： Unmanned aerial vehicles (UAV)

来源：评论

学校读者我要写书评

暂无评论

Model-Based Deep learning

引用

PROCEEDINGS OF THE ieee 2023年第5期111卷 465-499页

作者： Shlezinger, Nir Whang, Jay Eldar, Yonina C. Dimakis, Alexandros G. Ben Gurion Univ Negev Sch Elect & Comp Engn ECE IL-8410501 Beer Sheva Israel Univ Texas Austin Dept Comp Sci CS Austin TX 78712 USA Weizmann Inst Sci Fac Math & Comp Sci CS IL-7632706 Rehovot Israel Univ Texas Austin Dept Elect & Comp Engn ECE Austin TX 78712 USA

Signal processing, communications, and control have traditionally relied on classical statistical modeling techniques. Such model-based methods utilize mathematical formulations that represent the underlying physics, prior information, and additional domain knowledge. Simple classical models are useful but 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 that are model-agnostic are becoming increasingly popular as datasets become abundant and the power of modern deep learning pipelines increases. Deep neural networks (DNNs) use generic architectures that learn to operate from data and demonstrate excellent performance, especially for supervised problems. However, DNNs typically require massive amounts of data and immense computational resources, limiting their applicability for some scenarios. In this article, we present the leading approaches for studying and designing model-based deep learning systems. These are methods 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, and learning from limited data. Among the applications detailed in our examples for model-based deep learning are compressed sensing, digital communications, and tracking in state-space models. Our aim is to facilitate the design and study of future systems at the intersection of signal processing and machine learning that incorporate the advantages of both domains.

关键词： Deep learning Mathematical models Computational modeling data models Signal processing Signal processing algorithms Task analysis model-based machine learning signal processing

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：