检索结果-内蒙古大学图书馆

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Meng, Benyuan Xu, Qianqian Wang, Zitai Cao, Xiaochun Huang, Qingming Institute of Information Engineering CAS China School of Cyber Security University of Chinese Academy of Sciences China Key Lab. of Intelligent Information Processing Institute of Computing Technology CAS China Peng Cheng Laboratory China School of Cyber Science and Tech. Shenzhen Campus of Sun Yat-sen University China School of Computer Science and Tech. University of Chinese Academy of Sciences China Key Laboratory of Big Data Mining and Knowledge Management CAS China

Diffusion models are initially designed for image generation. Recent research shows that the internal signals within their backbones, named activations, can also serve as dense features for various discriminative tasks such as semantic segmentation. Given numerous activations, selecting a small yet effective subset poses a fundamental problem. To this end, the early study of this field performs a large-scale quantitative comparison of the discriminative ability of the activations. However, we find that many potential activations have not been evaluated, such as the queries and keys used to compute attention scores. Moreover, recent advancements in diffusion architectures bring many new activations, such as those within embedded ViT modules. Both combined, activation selection remains unresolved but overlooked. To tackle this issue, this paper takes a further step with a much broader range of activations evaluated. Considering the significant increase in activations, a full-scale quantitative comparison is no longer operational. Instead, we seek to understand the properties of these activations, such that the activations that are clearly inferior can be filtered out in advance via simple qualitative evaluation. After careful analysis, we discover three properties universal among diffusion models, enabling this study to go beyond specific models. On top of this, we present effective feature selection solutions for several popular diffusion models. Finally, the experiments across multiple discriminative tasks validate the superiority of our method over the SOTA competitors. Our code is available at this url. © 2024 Neural information processing systems foundation. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Research on Image Captioning Based on Vision-language Pre-trained Models 9

Research on Image Captioning Based on Vision-language Pre-tr...

引用

9th International Conference on Big data and Information Analytics, BigDIA 2023

作者： Zou, Shiwei Xie, Yuxiang Yan, Jie Wei, Yingmei Luan, Xidao National University of Defense Technology Department of System Engineering Hunan China National University of Defense Technology Key Laboratory of Big Data Department of System Engineering Changsha Hunan China Changsha University Department of Computer Science Hunan China

ISBN: (纸本)9798350330076

Image captioning is a vision-language task that targets at describing an image by generating a coherent sentence automatically. This technology allows computers to understand and describe images like humans, enabling further processing of images. It combines computer vision and natural language processing, which has broad applications. This paper investigates the use of vision-language pre-trained models for image captioning. In this article, an encoder-decoder hybrid model based on PVT and BERT is pre-trained on a large number of image-text pairs, possessing both understanding and generation capabilities. It consists of three components: 1) an image encoder and a text encoder to extract image and text features respectively;2) a multi-modal encoder to align the two modalities;and 3) a text decoder to generate text descriptions. This paper focuses on enhancing the image encoder of the model, which is crucial for image captioning models as it extracts high-resolution image features. Reducing training time and computational costs can also improve the applicability of pre-trained models. Improvements are made to the original model by modifying the ViT model architecture of the image encoding module. A feature pyramid structure and a spatial reduction attention mechanism is introduced to extract high-resolution image features while reducing computational complexity and memory usage, thereby enhancing the universality of the model. The model adopts a pure Transformer architecture, discarding the convolutional structure, and achieves excellent performance. © 2023 IEEE.

关键词： Image enhancement

来源：评论

学校读者我要写书评

暂无评论

Multi-Target Track Initiation in Heavy Clutter

引用

Computers, Materials & Continua 2022年第9期72卷 4489-4507页

作者： Li Xu Ruzhen Lou Chuanbin Zhang Bo Lang Weiyue Ding College of Computer Science and Technology Harbin Engineering UniversityHarbin150001China Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Jilin UniversityChangchun130012China Norinco Group Air Ammunition Research Institute Co.Ltd. Harbin150036China Harvard Medical School Boston02115USA

In the heavy clutter environment, the information capacity is large,the relationships among information are complicated, and track initiationoften has a high false alarm rate or missing alarm rate. Obviously, it is adifficult task to get a high-quality track initiation in the limited measurementcycles. This paper studies the multi-target track initiation in heavy *** first, a relaxed logic-based clutter filter algorithm is presented. In thealgorithm, the raw measurement is filtered by using the relaxed logic *** not only design a kind of incremental and adaptive filtering gate, but alsoadd the angle extrapolation based on polynomial extrapolation. The algorithm eliminates most of the clutter and obtains the environment with highdetection rate and less clutter. Then, we propose a fuzzy sequential Houghtransform-based track initiation algorithm. The algorithm establishes a newmeshing rule according to system noise to balance the relationship between thegrid granularity and the track initiation quality. And a flexible superpositionmatrix based on fuzzy clustering is constructed, which avoids the transformation error caused by 0–1 voting method in traditional Hough *** addition, the algorithm allows the superposition matrixes of nonadjacentcycles to be associated to overcome the shortcoming that the track can’t beinitiated in time when the measurements appear in an intermittent way. Anda slope verification method is introduced to detect formation-intensive serialtracks. Last, the sliding window method is employed to feedback the trackinitiation results timely and confirm the track. Simulation results verify thatthe proposed algorithms can initiate the tracks accurately in heavy clutter.

关键词： Track initiation heavy clutter multi-target Hough transform

来源：评论

学校读者我要写书评

暂无评论

Autonomous Human-Robot Collaborative Assembly Method Driven by the Fusion of Large Language Model and Digital Twin

Autonomous Human-Robot Collaborative Assembly Method Driven ...

引用

2024 International Conference on Intelligent Systems and Robotics, CISR 2024

作者： Chen, Jianpeng Luo, Haiwei Huang, Sihan Zhang, Meidi Wang, Guoxin Yan, Yan Jing, Shikai School of Mechanical Engineering Beijing Institute of Technology Beijing100081 China Capital Aerospace Machinery Corporation Limited Beijing100076 China Key Laboratory of Industry Knowledge & Data Fusion Technology and Application Ministry of Industry and Information Technology Beijing Institute of Technology Beijing100081 China China North Artificial Intelligence & Innovation Research Institute Beijing100071 China

Human-robot collaboration (HRC) plays an important role in human-centric manufacturing, which requires cooperative robots to have the ability of collaborate with human autonomously. It is very complex to understand the intention of human during the assembly process, therefore, we proposed a method of autonomous HRC assembly driven by the fusion of large language model (LLM) and digital twin in this paper. The assembly state is recognized from two perspectives, including the perception of key parts based on transfer learning and YOLO, and perceive operator actions based on LSTM and attention mechanism. In order to improve the autonomy of HRC, a collaborative task decision method driven by fine-tuning LLM based on assembly domain knowledge is proposed. A case study of reducer assembly is presented to verify the effectiveness of the proposed method. © Published under licence by IOP Publishing Ltd.

关键词： Collaborative robots

来源：评论

学校读者我要写书评

暂无评论

Bumpless Transfer Control of Switched Time-Delay Linear Systems with State-dependent Switching 41

Bumpless Transfer Control of Switched Time-Delay Linear Syst...

引用

第41届中国控制会议

作者： Jinghan Li Jun Zhao State Key Laboratory of Synthetical Automation for Process Industries Northeastern University Key Laboratory of Data Analytics and Optimization for Smart Industry (Northeastern University) Ministry of Education College of Information Science and Engineering Northeastern University

ISBN: (数字)9789887581536

ISBN: (纸本)9781665482561

This paper deals with the bumpless transfer control for switched time-delay *** time-delay bumpless transfer performance for switched linear system is provided,which can suppress control bumps caused by the current state and the time-delay state at the switching ***,by utilizing the multiple Lyapunov functions,under the co-design of the state-dependent switching law and the time-delay controller,a sufficient condition guaranteeing bumpless transfer control of switched time-delay systems is *** effectiveness of the result is validated by an *** result shows that our method reduces the bumpy amplitude while ensuring the stability of the switched time-delay system.

关键词： Switched Systems Time-delay State-dependent Bumpless Transfer

来源：评论

学校读者我要写书评

暂无评论

Breast Cancer Histopathology Image Classification Using Frequency Attention Convolution Network 9th

Breast Cancer Histopathology Image Classification Using Freq...

引用

19th International Conference on Advanced data Mining and Applications, ADMA 2023

作者： Lu, Ruidong Sun, Qiule Ding, Xueyan Zhang, Jianxin School of Computer Science and Engineering Dalian Minzu University Dalian China School of Information and Communication Engineering Dalian University of Technology Dalian China SEAC Key Laboratory of Big Data Applied Technology Dalian Minzu University Dalian China

ISBN: (纸本)9783031466632

The existing deep learning works mainly capture breast cancer histopathology image features in the spatial domain, and they rarely consider the frequency domain feature representation of histopathology images. According to the classical digital signal processing theory, frequency domain features may outperform spatial domain features in analyzing texture images. Motivated by this, we attempt to mine frequency domain features for the breast cancer histopathology image classification application, and further propose a novel frequency-attention convolutional network called SeFFT-Net by combining the Fourier transform with the channel attention mechanism. The core of SeFFT-Net consists of a newly constructed frequency-based squeeze and excitation (SeFFT) module, which first performs Fourier transform with residual construction to capture deep features in the frequency domain of histopathology images, followed by a squeeze-and-excitation attention operator to further enhance important frequency features. We extensively evaluate the proposed SeFFT-Net model on the public BreakHis breast cancer histopathology dataset, and it achieves the optimal image-level and patient-level classification accuracy of 98.67% and 98.16%, respectively. Meanwhile, ablation studies also well demonstrate the effectiveness of introducing frequency transforms for this medical image application. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Image classification

来源：评论

学校读者我要写书评

暂无评论

Spatial and temporal variation of water clarity in typical reservoirs in the Beijing-Tianjin-Hebei region observed by GF 1-WFV satellite data

引用

Journal of Oceanology and Limnology 2024年第4期42卷 1048-1060页

作者： Chang CAO Junsheng LI Xiaodong JIA Shenglei WANG Bo WAN Liaoning Earthquake Agency Shenyang 11000China Key Laboratory of Digital Earth Science Aerospace Information Research InstituteChinese Academy of SciencesBeijing 100094China International Research Center of Big Data for Sustainable Development Goals Beijing 100094China School of Electronic Electrical and Communication EngineeringUniversity of Chinese Academy of SciencesBeijing 100049China

Rapidly monitoring regional water quality and the changing trend is of great practical and scientific significance,especially for the Beijing-Tianjin-Hebei(BTH)region of China where water resources are relatively scarce and inland water bodies are generally *** remote sensing data of the GF 1 satellite launched in 2013 have characteristics of high spatial and temporal resolution,which can be used for the dynamic monitoring of the water environment in small lakes and ***,the water quality remote sensing monitoring model based on the GF 1 satellite data for lakes and reservoirs in BTH is still lacking because of the considerable differences in the optical characteristics of the lakes and *** this paper,the typical reservoirs in BTH-Guanting Reservoir,Yuqiao Reservoir,Panjiakou Reservoir,and Daheiting Reservoir are taken as the study *** the atmospheric correction of GF 1-WFV,the relative radiation normalized atmospheric correction was adopted after comparing it with other methods,such as 6 S and *** the water clarity retrieval,a water color hue angle based model was proposed and outperformed other available published models,with the R 2 of 0.74 and MRE of 31.7%.The clarity products of the four typical reservoirs in the BTH region in 2013-2019 were produced using the GF 1-WFV *** on the products,temporal and spatial changes in clarity were analyzed,and the main influencing factors for each water body were *** was found that the clarity of Guanting,Daheiting,and Panjiakou reservoirs showed an upward trend during this period,while that of Yuqiao Reservoir showed a downward *** the influencing factors,the water level of the water bodies can be an important factor related to the water clarity changes in this region.

关键词： GF 1 satellite atmospheric correction clarity Beijing-Tianjin-Hebei spatial and temporal change analysis

来源：评论

学校读者我要写书评

暂无评论

An Efficient Multilevel Threshold Image Segmentation Approach Based on Rafflesia Optimization Algorithm 1

引用

18th International Conference on Intelligent Information Hiding and Multimedia Signal Processing, IIH-MSP 2022

作者： Zhou, Jin-Liang Chu, Shu-Chuan Weng, Shaowei Zheng, Wei-Min College of Computer Science and Engineering Shandong University of Science and Technology Qingdao China College of Science and Engineering Flinders University Adelaide Australia Fujian Provincial Key Laboratory of Big Data Mining and Applications Fujian University of Technology Fuzhou China

ISBN: (数字)9789819901050

ISBN: (纸本)9789819901043

Segmentation is an important step from image processing to analysis. The thresholding technique based on entropy has become one of the most commonly used techniques in this field. Among the many thresholding techniques based on entropy, the minimum cross-entropy threshold (MCET) has been dramatically developed. The key to multilevel MCET technology is quickly finding the optimal threshold point. MCET is a relatively time-consuming technique, so we use optimization techniques to search for optimal multilevel thresholds. Rafflesia optimization algorithm (ROA) is an optimization algorithm inspired by rafflesia growth theory to solve real-world multi-parameter optimization problems. In this paper, Rafflesia optimization algorithm, particle swarm optimization, gray wolf optimization, whale optimization algorithm, and multi-version optimizer are used to segment six images and the peak signal-to-noise ratio (PSNR), structural similarity index (SSIM), and feature similarity index (FSIM) as evaluation metrics. The experimental results show that the ROA algorithm can achieve higher-quality image segmentation. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

关键词： Image segmentation

来源：评论

学校读者我要写书评

暂无评论

An efficient image encryption scheme based on a novel hyperchaotic 2D-ACSES model and bit-level folding permutation

引用

The European Physical Journal Plus 2025年第6期140卷 1-21页

作者： Li, Zhen Zhang, Shuang Tan, Weijie Wu, Xianming College of Big Data and Information Engineering Guizhou University Guiyang China State Key Laboratory of Public Big Data Guizhou University Guiyang China College of Big Data and Information Engineering Guiyang Institute of Humanities and Technology Guiyang China Guizhou Big Data Academy Guizhou University Guiyang China School of Mechanical and Electrical Engineering Guizhou Normal University Guiyang China

Image encryption using chaotic maps is an important approach for securing digital images in the era of widespread Internet communication. We propose a novel discrete chaotic system, the two-dimensional Arcsin-Cubic-Sine-Exp System (2D-ACSES), which exhibits hyperchaotic dynamics, making it suitable for cryptographic applications. Based on the 2D-ACSES system, we introduce a new encryption scheme using bit-level folding permutation, improving the permutation efficiency. The process begins with an internal key generation method, transforming both the plaintext image and the initial key into internal keys, which initialize the system. The image matrix is reshaped into a bit matrix, which undergoes several transformation stages, including folding, row shifting, and refolding to increase confusion. Finally, the modified bit matrix is restored to the image matrix, which is further encrypted through row diffusion and pixel-level permutation. These steps enhance the diffusion of information, ensuring that the ciphertext is highly sensitive to both the plaintext and the initial keys, thus providing robust security. Simulation results and analysis demonstrate that the proposed encryption scheme provides strong security performance.

关键词：

来源：评论

学校读者我要写书评

暂无评论

TransD-based Multi-hop Meta Learning for Few-shot knowledge Graph Completion

TransD-based Multi-hop Meta Learning for Few-shot Knowledge ...

引用

International Joint Conference on Neural Networks (IJCNN)

作者： Jindi Li Kui Yu Yuling Li Yuhong Zhang Key Laboratory of Knowledge Engineering with Big Data of Ministry of Education Hefei University of Technology Hefei China School of Computer Science and Information Engineering Hefei University of Technology Hefei China

Few-shot knowledge graph completion (FKGC), which aims to infer missing facts about a relation from only a few reference triples, has recently attracted great attention. The core of solving the FKGC task is to learn a vector representation for each few-shot relation using the corresponding entity represen-tations. To this end, existing models generally enhance entity representations with their direct neighbors. However, a large number of entities have few direct neighbors. Hence, encoding only direct neighborhood is insufficient to obtain satisfactory en-tity representations. In addition, current models typically utilize static embeddings to represent entities, ignoring their diverse semantics, i.e., an entity may show distinct semantics within different few-shot relations. To address these issues, we propose a new FKGC framework, namely TransD-based Multi-hop Meta Learning (TDML). TDML consists of three main components: a multi-hop neighbor encoder to enhance entity representations by aggregating heterogeneous multi-hop neighbors, a transformer encoder to generate the relation meta representations, and a TransD-based relation representation updater that allows each entity to exhibit relation-specific semantics and tune the relation meta representations. Extensive experiments on two public datasets demonstrate that our model outperforms state-of-the-art FKGC methods.

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：