检索结果-内蒙古大学图书馆

Category re-balancing based domain adaptive semantic segmentation

Procedia computer science 2024年 242卷 1155-1162页

作者： Siyu Zhu Fenfen Zhou Kai Li Minghao Liu Haonan Wen Computer School Beijing information science and technology university 102206 No.55 Taihang Road Changping District Beijing China Beijing Key Laboratory of Information Service Engineering Beijing Union University 100101 Beijing China. College of Robotics Beijing Union University 100027 Beijing China National Engineering Research Center of Visual Technology School of Computer Science Peking University 100871 Beijing China School of Computer Science and Technology University of Chinese Academy of Sciences No.19(A) Yuquan Road Shijingshan District Beijing 100049 P.R. China Research Center on Fictitious Economy and Data Science Chinese Academy of Sciences No. 80 of Zhongguancun East Road Haidian District Beijing 100190 Beijing China

The semantic segmentation task faces the bottleneck of high manual annotation costs. Domain adaptive learning provides an effective solution through inter domain knowledge transfer. However, existing domain adaptive semantic segmentation tasks face the situation of category asymmetry between scenes when dealing with complex scenes. Existing methods lack handling of class imbalance in the prediction process. In domain adaptive semantic segmentation tasks, class imbalance limits the performance of inter domain transfer. Our work focuses on the impact of category distribution on domain adaptive semantic segmentation tasks in image pairs. Inspired by long tail learning, we divided categories into head, middle, and tail categories, and designed effective strategies to enhance the contribution of tail categories in domain adaptation tasks. We introduce the concept of category rebalancing into domain adaptive image matching, and our method is lightweight and effective, proving its effectiveness on general datasets.

关键词： Long-tail learning Domain adaptation Semantic segmentation Transfer learning

来源：评论

学校读者我要写书评

暂无评论

Iterative Joint Detection-Decoding Algorithms Using Euclidean Distance-Based Feedback 10

Iterative Joint Detection-Decoding Algorithms Using Euclidea...

引用

10th International Conference on Information and Automation for Sustainability, ICIAfS 2021

作者： Ullah, Waheed Jayakody, Dushantha Nalin K. Li, Jun Chursin, Yuri University of the Witwatersrand School of Electrical and Information Egineering Johannesburg South Africa Nanjing University of Science and Technology School of Electronic and Optical Engineering Nanjing210094 China Tomsk Polytechnic University School of Computer Science and Robotics Tomsk634050 Russia

ISBN: (纸本)9781665444293

In this article, two algorithms for improving the bit error rate performance of the non-binary LDPC codes over higher order modulation are proposed. The first algorithm is the predictive syndrome based symbol flipping decoding algorithm. The flipping function of this algorithm utilizes the channel reliability to identify the least reliable symbol position. In Algorithm 1, if the predicted symbol value satisfies the check sum, then the candidate symbol value is declared as correct otherwise that value is adjusted and sent back to the QAM detector. Algorithm 2 in this paper is an improvement to iterative joint detection-decoding algorithm by using the method of iterative hard decision based majority logic to select the new candidate symbol value. The feedback value to the QAM detector is adjusted by using Euclidean distance between the current symbol and the newly selected symbol value. Numerical results and complexity analysis show that the proposed schemes have better bit error rate versus complexity trade-off in comparison to some of the existing algorithms. © 2021 IEEE.

关键词： Bit error rate

来源：评论

学校读者我要写书评

暂无评论

Adaptive Multi-modal Fusion of Spatially Variant Kernel Refinement with Diffusion Model for Blind Image Super-Resolution

arXiv

引用

arXiv 2024年

作者： Lin, Junxiong Wang, Yan Tao, Zeng Wang, Boyang Zhao, Qing Wang, Haorang Tong, Xuan Mai, Xinji Lin, Yuxuan Song, Wei Yu, Jiawen Yan, Shaoqi Zhang, Wenqiang Shanghai Engineering Research Center of AI & Robotics Academy for Engineering & Technology Fudan University Shanghai China East China University of Science and Technology Shanghai China College of Information Technology Shanghai Ocean University Shanghai China Engineering Research Center of AI & Robotics Ministry of Education Academy for Engineering & Technology Fudan University Shanghai China Shanghai Key Lab of Intelligent Information Processing School of Computer Science Fudan University Shanghai China

Pre-trained diffusion models utilized for image generation encapsulate a substantial reservoir of a priori knowledge pertaining to intricate textures. Harnessing the potential of leveraging this a priori knowledge in the context of image super-resolution presents a compelling avenue. Nonetheless, prevailing diffusion-based methodologies presently overlook the constraints imposed by degradation information on the diffusion process. Furthermore, these methods fail to consider the spatial variability inherent in the estimated blur kernel, stemming from factors such as motion jitter and out-of-focus elements in open-environment scenarios. This oversight results in a notable deviation of the image super-resolution effect from fundamental realities. To address these concerns, we introduce a framework known as Adaptive Multi-modal Fusion of Spatially Variant Kernel Refinement with Diffusion Model for Blind Image Super-Resolution (SSR). Within the SSR framework, we propose a Spatially Variant Kernel Refinement (SVKR) module. SVKR estimates a Depth-Informed Kernel, which takes the depth information into account and is spatially variant. Additionally, SVKR enhance the accuracy of depth information acquired from LR images, allowing for mutual enhancement between the depth map and blur kernel estimates. Finally, we introduce the Adaptive Multi-Modal Fusion (AMF) module to align the information from three modalities: low-resolution images, depth maps, and blur kernels. This alignment can constrain the diffusion model to generate more authentic SR results. Copyright © 2024, The Authors. All rights reserved.

关键词： Image enhancement

来源：评论

学校读者我要写书评

暂无评论

Robotic Control Optimization Through Kernel Selection in Safe Bayesian Optimization

Robotic Control Optimization Through Kernel Selection in Saf...

引用

IEEE International Conference on robotics and Biomimetics

作者： Lihao Zheng Hongxuan Wang Xiaocong Li Jun Ma Prahlad Vadakkepat School of Data Science The Chinese University of Hong Kong Research Intern at the Singapore Institute of Manufacturing Technology A*STAR Shenzhen China Department of Electrical and Computer Engineering National University of Singapore Singapore Singapore Institute of Manufacturing Technology Agency for Science Technology and Research (A*STAR) Singapore Robotics and Autonomous Systems Thrust The Hong Kong University of Science and Technology (Guangzhou) Guangzhou China

ISBN: (数字)9798331509644

ISBN: (纸本)9798331509651

Control system optimization has long been a fundamental challenge in robotics. While recent advancements have led to the development of control algorithms that leverage learning-based approaches, such as SafeOpt, to optimize single feedback controllers, scaling these methods to high-dimensional complex systems with multiple controllers remains an open problem. In this paper, we propose a novel learning-based control optimization method, which enhances the additive Gaussian process-based Safe Bayesian Optimization algorithm to efficiently tackle high-dimensional problems through kernel selection. We use PID controller optimization in drones as a representative example and test the method on Safe Control Gym, a benchmark designed for evaluating safe control techniques. We show that the proposed method provides a more efficient and optimal solution for high-dimensional control optimization problems, demonstrating significant improvements over existing techniques.

关键词： Additives Trajectory tracking Optimization methods Control systems Bayes methods Kernel Robots Tuning Standards Quadrotors

来源：评论

学校读者我要写书评

暂无评论

SSD-MonoDETR: Supervised Scale-aware Deformable Transformer for Monocular 3D Object Detection

arXiv

引用

arXiv 2023年

作者： He, Xuan Yang, Fan Yang, Kailun Lin, Jiacheng Fu, Haolong Wang, Meng Yuan, Jin Li, Zhiyong The College of Computer Science and Electronic Engineering Hunan University Changsha410082 China The School of Robotics Hunan University China410012 China The National Engineering Research Center of Robot Visual Perception and Control Technology Hunan University Changsha410082 China The School of Computer Science Hefei University of Technology Hefei230009 China

Transformer-based methods have demonstrated superior performance for monocular 3D object detection recently, which aims at predicting 3D attributes from a single 2D image. Most existing transformer-based methods leverage both visual and depth representations to explore valuable query points on objects, and the quality of the learned query points has a great impact on detection accuracy. Unfortunately, existing unsupervised attention mechanisms in transformers are prone to generate low-quality query features due to inaccurate receptive fields, especially on hard objects. To tackle this problem, this paper proposes a novel "Supervised Scale-aware Deformable Attention" (SSDA) for monocular 3D object detection. Specifically, SSDA presets several masks with different scales and utilizes depth and visual features to adaptively learn a scale-aware filter for object query augmentation. Imposing the scale awareness, SSDA could well predict the accurate receptive field of an object query to support robust query feature generation. Aside from this, SSDA is assigned with a Weighted Scale Matching (WSM) loss to supervise scale prediction, which presents more confident results as compared to the unsupervised attention mechanisms. Extensive experiments on the KITTI and Waymo Open datasets demonstrate that SSDA significantly improves the detection accuracy, especially on moderate and hard objects, yielding state-of-the-art performance as compared to the existing approaches. Our code will be made publicly available at https://***/mikasa3lili/SSD-MonoDETR. Copyright © 2023, The Authors. All rights reserved.

关键词： Object detection

来源：评论

学校读者我要写书评

暂无评论

Improving Trust Estimation in Human-Robot Collaboration Using Beta Reputation at Fine-grained Timescales

arXiv

引用

arXiv 2024年

作者： Dagdanov, Resul Andrejević, Milan Liu, Dikai Lin, Chin-Teng The Robotics Institute Faculty of Engineering and Information Technology University of Technology Sydney UltimoNSW2007 Australia The Psychology Discipline Graduate School of Health Faculty of Health University of Technology Sydney UltimoNSW2007 Australia The Australian Artificial Intelligence Institute School of Computer Science Faculty of Engineering and Information Technology University of Technology Sydney UltimoNSW2007 Australia

When interacting with each other, humans adjust their behavior based on perceived trust. However, to achieve similar adaptability, robots must accurately estimate human trust at sufficiently granular timescales during the human-robot collaboration task. A beta reputation is a popular way to formalize a mathematical estimation of human trust. However, it relies on binary performance, which updates trust estimations only after each task concludes. Additionally, manually crafting a reward function is the usual method of building a performance indicator, which is labor-intensive and time-consuming. These limitations prevent efficiently capturing continuous changes in trust at more granular timescales throughout the collaboration task. Therefore, this paper presents a new framework for the estimation of human trust using a beta reputation at fine-grained timescales. To achieve granularity in beta reputation, we utilize continuous reward values to update trust estimations at each timestep of a task. We construct a continuous reward function using maximum entropy optimization to eliminate the need for the laborious specification of a performance indicator. The proposed framework improves trust estimations by increasing accuracy, eliminating the need for manually crafting a reward function, and advancing toward developing more intelligent robots. The source code is publicly available. © 2024, CC BY.

关键词： Intelligent robots

来源：评论

学校读者我要写书评

暂无评论

Byzantine-robust Decentralized Federated Learning via Dual-domain Clustering and Trust Bootstrapping

Byzantine-robust Decentralized Federated Learning via Dual-d...

引用

Conference on computer Vision and Pattern Recognition (CVPR)

作者： Peng Sun Xinyang Liu Zhibo Wang Bo Liu College of Computer Science and Electronic Engineering Hunan University China Department of Aeronautical and Aviation Engineering The Hong Kong Polytechnic University China School of Cyber Science and Technology Zhejiang University China Shenzhen Institute of Artificial Intelligence and Robotics for Society (AIRS) China

ISBN: (数字)9798350353006

ISBN: (纸本)9798350353013

Decentralized federated learning (DFL) facilitates collaborative model training across multiple connected clients without a central coordination server, thereby avoiding the single point of failure in traditional centralized federated learning (CFL). However, DFL exhibits increased susceptibility to Byzantine attacks owing to the lack of a responsible central server. Furthermore, a benign client in DFL may be dominated by Byzantine clients (more than half of its neighbors are malicious), posing significant challenges for robust model training. In this work, we propose DFL-Dual, a novel Byzantine-robust DFL method through dual-domain client clustering and trust bootstrapping. Specifically, we first propose to leverage both data-domain and model-domain distance metrics to identify client discrepancies. Then, we design a trust evaluation mechanism centered on benign clients, which enables them to evaluate their neighbors. Building upon the dual-domain distance metric and trust evaluation mechanism, we further develop a two-stage clustering and trust bootstrapping technique to exclude Byzantine clients from local model aggregation. We extensively evaluate the proposed DFL-Dual method through rigorous experimentation, demonstrating its remarkable performance superiority over existing robust CFL and DFL schemes.

关键词： Measurement Training computer vision Federated learning Buildings Collaboration Euclidean distance

来源：评论

学校读者我要写书评

暂无评论

A Stable Lithium-ion Battery SOH Estimation Framework for Suppressing Measurement Noise with Unknown Distribution

引用

IEEE Transactions on Transportation Electrification 2025年

作者： Ma, Wentao Xue, Jingsong Li, Yang Guo, Peng Liu, Xinghua Wei, Zhongbao Wang, Yiwen Chen, Badong Xi’an University of Technology School of Electrical Engineering 710048 China Chalmers University of Technology Department of Electrical Engineering Gothenburg41296 Sweden Beijing Institute of Technology National Engineering Laboratory for Electric Vehicles School of Mechanical Engineering Beijing100081 China Hong Kong University of Science and Technology Department of Electronic and Computer Engineering Hong Kong Institute of artificial intelligence and robotics Xi’an Jiaotong University Xi’an710048 China

Existing methods for estimating the state of health (SOH) of lithium-ion battery (LIB) typically rely on the assumption that the distribution of noise (or outliers) in the measurement data is known. However, this assumption rarely holds true for LIB operating under real-word conditions. This article proposes a stable framework for accurate SOH estimation that accommodates noises with unknown distribution in both measurement data and label values. The framework combines generalized correntropy loss (GCL) with Savitzky-Golay (SG) filter and extreme learning machine (ELM) to obtain measurement data filter named SG-GCL and SOH estimator named generalized ELM (GELM), respectively. The SG-GCL filtering of the measurement data keeps the root mean square error (RMSE) within 0.0365%, and Pearson correlation between extracted feature and SOH improves by 0.4963, which in turn leads to the reduction of the RMSE metrics of the ELM for the estimation of the SOH by 43.69%. From the filtering results, feature extraction and estimation results proved its necessity and effectiveness. GELM effectively suppresses the influence of label value noise on the model in the training process, which reduces the SOH estimation RMSE index by more than 0.66%. The results from experiments with different distributional noise conditions show that the proposed SOH estimation framework has excellent and stable performance. © 2015 IEEE.

关键词： Mean square error

来源：评论

学校读者我要写书评

暂无评论

Guidewire Endpoint Detection Based on Pixel Adjacent Relation in Robot-assisted Cardiovascular Interventions

Guidewire Endpoint Detection Based on Pixel Adjacent Relatio...

引用

Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC)

作者： Wenjing Du Guanlin Yi Olatunji Mumini Omisore Wenke Duan Toluwanimi Oluwadara Akinyemi Xingyu Chen Lei Wang Boon-Giin Lee Jiang Liu Research Centre for Medical Robotics and Minimally Invasive Surgical Devices Shenzhen Institutes of Advanced Technology Chinese Academy of Sciences Shenzhen China School of Computer Science University of Nottingham Ningbo China Ningbo China Academy for Engineering and Technology Fudan University Shanghai China Department of Computer Science and Engineering Southern University of Science and Technology Shenzhen China

Visualization of endovascular tools like guidewire and catheter is essential for procedural success of endovascular interventions. This requires tracking the tool pixels and motion during catheterization; however, detecting the endpoints of the endovascular tools is challenging due to their small size, thin appearance, and flexibility. As this still limit the performances of existing methods used for endovascular tool segmentation, predicting correct object location could provide ways forward. In this paper, we proposed a neighborhood-based method for detecting guidewire endpoints in X-ray angiograms. Typically, it consists of pixel-level segmentation and a post-segmentation step that is based on adjacency relationships of pixels in a given neighborhood. The latter includes skeletonization to predict endpoint pixels of guidewire. The method is evaluated with proprietary guidewire dataset obtained during in-vivo study in six rabbits, and it shows a high segmentation performance characterized with precision of 87.87% and recall of 90.53%, and low detection error with a mean pixel error of 2.26±0.14 pixels. We compared our method with four state-of-the-art detection methods and found it to exhibit the best detection performance. This neighborhood-based detection method can be generalized for other surgical tool detection and in related computer vision *** Relevance— The proposed method can be provided with better tool tracking and visualization systems during robot-assisted intravascular interventional surgery.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Incentive Mechanism Design for Distributed Ensemble Learning

Incentive Mechanism Design for Distributed Ensemble Learning

引用

IEEE Conference on Global Communications (GLOBECOM)

作者： Chao Huang Pengchao Han Jianwei Huang Department of Computer Science University of California Davis USA School of Information Engineering Guangdong University of Technology China School of Science and Engineering The Chinese University of Hong Kong Shenzhen China Shenzhen Institute of Artificial Intelligence and Robotics for Society China

Distributed ensemble learning (DEL) involves training multiple models at distributed learners, and then combining their predictions to improve performance. Existing related studies focus on algorithm development but ignore the important issue of incentives, without which self-interested learners may be unwilling to participate. We aim to fill this gap by presenting a first study on the incentive mechanism design in DEL. The mechanism specifies both the training data and the reward for learners with heterogeneous computation and communication costs. One challenge is that it is unclear how learners' diversity (in terms of training data) contributes to the ensemble accuracy. To this end, we decompose the ensemble accuracy into a diversity-precision tradeoff to guide the mechanism design. Another challenge is that the mechanism design is a mixed-integer program with a large search space. To this end, we propose an alternating algorithm that iteratively updates each learner's training data size and reward. We prove that the algorithm converges and is polynomial in the number of learners. Numerical results using MNIST dataset are consistent with our analysis. Interestingly, we show that the mechanism may prefer a lower level of learner diversity to achieve a higher ensemble accuracy. Our code is made publicly available.

关键词：

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：