检索结果-内蒙古大学图书馆

2024 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2024

作者： Zhou, Xiaolong Yuan, Qianmu Zheng, Shuangjia Wang, Yu Yang, Yuedong Sun Yat-sen University Pengcheng Laboratory School of Computer Science and Engineering Guangzhou China Sun Yat-sen University National Supercomputing Center in Shenzhen School of Computer Science and Engineering Guangzhou China China Peng Cheng Laboratory Shenzhen China Sun Yat-sen University School of Computer Science and Engineering Guangzhou China

ISBN: (纸本)9798350386226

Nanobodies are emerging therapeutic antibodies with more simple structure, which can target antigen surfaces and tissue types not accessible to conventional antibodies. However, nanobodies exhibit polyreactivity, binding non-specifically to off-target proteins and other biomolecules. This uncertainty can affect the drug development process and pose significant challenges in clinical development. Existing computational polyreactivity prediction methods focus solely on the sequence or fail to fully utilize structural information. In this study, we propose GP-nano, a geometric graph network based model for nanobody polyreactivity using predictive structure. GP-nano starts from sequences, predicts protein structures using ESMfold, and fully utilizes structural geometric information via graph networks. GP-nano can accurately classify the polyreactivity of nanobodies (AUC=0.91). To demonstrate GP-nano's generalizability, we also trained and tested it on monoclonal antibodies (mAbs). GP-nano outperforms the best methods on both datasets, indicating the contribution of structural information and geometric features to antibody polyreactivity prediction. © 2024 IEEE.

关键词： Nanoclay

来源：评论

学校读者我要写书评

暂无评论

MICM: Rethinking Unsupervised Pretraining for Enhanced Few-shot Learning 24

MICM: Rethinking Unsupervised Pretraining for Enhanced Few-s...

引用

32nd ACM International Conference on Multimedia, MM 2024

作者： Zhang, Zhenyu Chen, Guangyao Zou, Yixiong Huang, Zhimeng Li, Yuhua Li, Ruixuan School of Computer Science and Technology Huazhong University of Science and Technology Wuhan China National Key Laboratory for Multimedia Information Processing School of Computer Science Peking University Beijing China National Engineering Research Center of Visual Technology School of Computer Science Peking University Beijing China

ISBN: (纸本)9798400706868

Humans exhibit a remarkable ability to learn quickly from a limited number of labeled samples, a capability that starkly contrasts with that of current machine learning systems. Unsupervised Few-Shot Learning (U-FSL) seeks to bridge this divide by reducing reliance on annotated datasets during initial training phases. In this work, we first quantitatively assess the impacts of Masked Image Modeling (MIM) and Contrastive Learning (CL) on few-shot learning tasks. Our findings highlight the respective limitations of MIM and CL in terms of discriminative and generalization abilities, which contribute to their underperformance in U-FSL contexts. To address these trade-offs between generalization and discriminability in unsupervised pretraining, we introduce a novel paradigm named Masked Image Contrastive Modeling (MICM). MICM creatively combines the targeted object learning strength of CL with the generalized visual feature learning capability of MIM, significantly enhancing its efficacy in downstream few-shot learning inference. Extensive experimental analyses confirm the advantages of MICM, demonstrating significant improvements in both generalization and discrimination capabilities for few-shot learning. Our comprehensive quantitative evaluations further substantiate the superiority of MICM, showing that our two-stage U-FSL framework based on MICM markedly outperforms existing leading baselines. The repository of this project is available at https://***/iCGY96/MICM. © 2024 ACM.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Prediction of the Pore-Pressure Built-Up and Temperature of Fire-Loaded Concrete with Pix2Pix

引用

computers, Materials & Continua 2024年第5期79卷 2907-2922页

作者： Xueya Wang Yiming Zhang Qi Liu Huanran Wang Mechanics and Materials Science Research Center Ningbo UniversityNingbo315211China Jinyun Institute Zhejiang Sci-Tech UniversityLishui321400China School of Civil Engineering and Architecture Zhejiang Sci-Tech UniversityHangzhou310018China School of Computer Science Nanjing University of Information Science and TechnologyNanjing210044China

Concrete subjected to fire loads is susceptible to explosive spalling, which can lead to the exposure of reinforcingsteel bars to the fire, substantially jeopardizing the structural safety and stability. The spalling of fire-loaded concreteis closely related to the evolution of pore pressure and temperature. Conventional analytical methods involve theresolution of complex, strongly coupled multifield equations, necessitating significant computational efforts. Torapidly and accurately obtain the distributions of pore-pressure and temperature, the Pix2Pix model is adoptedin this work, which is celebrated for its capabilities in image generation. The open-source dataset used hereinfeatures RGB images we generated using a sophisticated coupled model, while the grayscale images encapsulate the15 principal variables influencing spalling. After conducting a series of tests with different layers configurations,activation functions and loss functions, the Pix2Pix model suitable for assessing the spalling risk of fire-loadedconcrete has been meticulously designed and trained. The applicability and reliability of the Pix2Pix model inconcrete parameter prediction are verified by comparing its outcomes with those derived fromthe strong couplingTHC model. Notably, for the practical engineering applications, our findings indicate that utilizing monochromeimages as the initial target for analysis yields more dependable results. This work not only offers valuable insightsfor civil engineers specializing in concrete structures but also establishes a robust methodological approach forresearchers seeking to create similar predictive models.

关键词： Fire loaded concrete spalling risk pore pressure generative adversarial network(GAN) Pix2Pix

来源：评论

学校读者我要写书评

暂无评论

Zero-Shot Event-Intensity Asymmetric Stereo via Visual Prompting from Image Domain 38

Zero-Shot Event-Intensity Asymmetric Stereo via Visual Promp...

引用

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Lou, Hanyue Liang, Jinxiu Teng, Minggui Fan, Bin Xu, Yong Shi, Boxin State Key Laboratory of Multimedia Information Processing School of Computer Science Peking University China National Engineering Research Center of Visual Technology School of Computer Science Peking University China National Key Laboratory of General AI School of Intelligence Science and Technology Peking University China School of Computer Science and Engineering South China University of Technology China

Event-intensity asymmetric stereo systems have emerged as a promising approach for robust 3D perception in dynamic and challenging environments by integrating event cameras with frame-based sensors in different views. However, existing methods often suffer from overfitting and poor generalization due to limited dataset sizes and lack of scene diversity in the event domain. To address these issues, we propose a zero-shot framework that utilizes monocular depth estimation and stereo matching models pretrained on diverse image datasets. Our approach introduces a visual prompting technique to align the representations of frames and events, allowing the use of off-the-shelf stereo models without additional training. Furthermore, we introduce a monocular cue-guided disparity refinement module to improve robustness across static and dynamic regions by incorporating monocular depth information from foundation models. Extensive experiments on real-world datasets demonstrate the superior zero-shot evaluation performance and enhanced generalization ability of our method compared to existing approaches. © 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

Evaluation of Emotion Recognition in Facial Image Using Attentive Deep Convolutional Neural Network 2

Evaluation of Emotion Recognition in Facial Image Using Atte...

引用

2nd International Conference on Information Technology Research and Innovation, ICITRI 2023

作者： Aini, Nur Pardede, Hilman Ferdinandus Graduate School of Computer Science Universitas Nusa Mandiri Faculty of Technology Information Depok Indonesia Research Center for AI and Cybersecurity National Research and Innovation Agency Indonesia

ISBN: (纸本)9798350324945

Emotion recognition in facial images is a topic that has attracted many interests. Research on emotion recognition in facial images continues to face challenges such as variations of human faces due to environments or situations. Of all the techniques that have been used, the Deep Convolution Neural Network (DCNN) method reports a significant improvement compared to traditional machine learning methods. In recent years, self-attachment mechanisms have increased in popularity in DCNN, especially in Natural Language Processing. In this study, we propose to add attention networks to DCNN method to improve the performance of emotion recognition. By incorporating self-attention mechanisms into the model architecture, the network is able to recognize which facial features are most relevant for each emotion, increasing the accuracy and interpretation of the model. This study reports an accuracy of 69.55% which is 11.15% higher than using the traditional DCNN model. © 2023 IEEE.

关键词： Emotion Recognition

来源：评论

学校读者我要写书评

暂无评论

Towards optimized tensor code generation for deep learning on sunway many-core processor

引用

Frontiers of computer science 2024年第2期18卷 1-15页

作者： Mingzhen LI Changxi LIU Jianjin LIAO Xuegui ZHENG Hailong YANG Rujun SUN Jun XU Lin GAN Guangwen YANG Zhongzhi LUAN Depei QIAN State Key Laboratory of Software Development Environment Beijing 100191China School of Computer Science and Engineering Beihang UniversityBeijing 100191China National University of Singapore Singapore 119077Singapore State Key Laboratory of Mathematical Engineering and Advanced Computing Wuxi 214000China Science and Technology on Special System Simulation Laboratory Beijing Simulation Center Beijing 100854China Department of Computer Science and Technology Tsinghua UniversityBeijing 100084China

The flourish of deep learning frameworks and hardware platforms has been demanding an efficient compiler that can shield the diversity in both software and hardware in order to provide application *** the existing deep learning compilers,TVM is well known for its efficiency in code generation and optimization across diverse hardware *** the meanwhile,the Sunway many-core processor renders itself as a competitive candidate for its attractive computational power in both scientific computing and deep learning *** paper combines the trends in these two ***,we propose swTVM that extends the original TVM to support ahead-of-time compilation for architecture requiring cross-compilation such as *** addition,we leverage the architecture features during the compilation such as core group for massive parallelism,DMA for high bandwidth memory transfer and local device memory for data locality,in order to generate efficient codes for deep learning workloads on *** experiment results show that the codes generated by swTVM achieve 1.79x improvement of inference latency on average compared to the state-of-the-art deep learning framework on Sunway,across eight representative *** work is the first attempt from the compiler perspective to bridge the gap of deep learning and Sunway processor particularly with productivity and efficiency in *** believe this work will encourage more people to embrace the power of deep learning and Sunwaymany-coreprocessor.

关键词： sunway processor deep learning compiler code generation performance optimization

来源：评论

学校读者我要写书评

暂无评论

Enhancing Video Anomaly Detection Using Spatio-Temporal Autoencoders and Convolutional LSTM Networks

引用

SN computer science 2024年第1期5卷 190页

作者： Almahadin, Ghayth Subburaj, Maheswari Hiari, Mohammad Sathasivam Singaram, Saranya Kolla, Bhanu Prakash Dadheech, Pankaj Vibhute, Amol D. Sengan, Sudhakar Department of Networks and Cybersecurity Faculty of Information Technology Al Ahliyya Amman University Country Amman Jordan School of Computer Science and Engineering Vellore Institute of Technology Tamil Nadu Chennai 600127 India Department of Networks and Cybersecurity Information Technology Al Ahliyya Amman University Amman Jordan Department Computing Technology School of Computing SRM Institute of Science and Technology Kattankulathur Campus Tamil Nadu Chennai 603203 India Department of Computer Science and Engineering Koneru Lakshmaiah Education Foundation Vaddeswaram Guntur 522302 India Department of Computer Science and Engineering Swami Keshvanand Institute of Technology Management and Gramothan (SKIT) Rajasthan Jaipur 302017 India Symbiosis Institute of Computer Studies and Research (SICSR) Symbiosis International (Deemed University) MH Pune 411016 India Department of Computer Science and Engineering PSN College of Engineering and Technology Tamil Nadu Tirunelveli 627152 India

Identifying suspicious activities or behaviors is essential in the domain of Anomaly Detection (AD). In crowded scenes, the presence of inter-object occlusions often complicates the detection of such behaviors. Therefore, developing a robust method capable of accurately detecting and locating anomalous activities within video sequences becomes crucial, especially in densely populated environments. This research initiative aims to address this challenge by proposing a novel approach focusing on AD behaviors in crowded settings. By leveraging a spatio-temporal method, the proposed approach harnesses the power of both spatial and temporal dimensions. This enables the method to effectively capture and analyze the intricate motion patterns and spatial information embedded within the continuous frames of video data. The objective is to create a comprehensive model that can efficiently detect and precisely locate anomalies within complex video sequences, specifically those featuring human crowds. The efficacy of the proposed model will be rigorously evaluated using a benchmark dataset encompassing diverse scenarios involving crowded environments. The dataset is designed to simulate real-world conditions where millions of video footage need to be continuously monitored in real time. The focus is on identifying anomalies, which might occur within short time frames, sometimes as brief as five minutes or even less. Given the challenges posed by the massive volume of data and the requirement for rapid AD, the research emphasizes the limitations of traditional Supervised Learning (SL) methods in this context. © 2024, The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd.

关键词： Anomaly detection Autoencoders LSTM Machine learning Spatio-temporal features Suspicious activities

来源：评论

学校读者我要写书评

暂无评论

Hyperspectral Image Classification Based on A Multi-Scale Weighted Kernel Network

引用

Chinese Journal of Electronics 2022年第5期31卷 832-843页

作者： SUN Le XU Bin LU Zhenyu School of Computer Science Nanjing University of Information Science and Technology Engineering Research Center of Digital Forensics Ministry of Education Nanjing University of Information Science and Technology School of Artificial Intelligence Nanjing University of Information Science and Technology

Recently, many deep learning models have shown excellent performance in hyperspectral image(HSI) classification. Among them, networks with multiple convolution kernels of different sizes have been proved to achieve richer receptive fields and extract more representative features than those with a single convolution kernel. However, in most networks, different-sized convolution kernels are usually used directly on multibranch structures, and the image features extracted from them are fused directly and simply. In this paper, to fully and adaptively explore the multiscale information in both spectral and spatial domains of HSI, a novel multi-scale weighted kernel network(MSWKNet) based on an adaptive receptive field is proposed. First, the original HSI cubic patches are transformed to the input features by combining the principal component analysis and one-dimensional spectral convolution. Then, a three-branch network with different convolution kernels is designed to convolve the input features, and adaptively adjust the size of the receptive field through the attention mechanism of each branch. Finally, the features extracted from each branch are fused together for the task of *** on three well-known hyperspectral data sets show that MSWKNet outperforms many deep learning networks in HSI classification.

关键词： multibranch structures image features convolution kernels multiple convolution kernels adaptive receptive field image classification feature extraction HSI classification single convolution kernel convolutional neural nets hyperspectral image classification representative features hyperspectral imaging geophysical image processing deep learning (artificial intelligence) multiscale weighted kernel network deep learning networks hyperspectral data sets three-branch network HSI cubic patches principal component analysis 1D spectral convolution different-sized convolution kernels

来源：评论

学校读者我要写书评

暂无评论

Ultrafast magneto-optical dynamics in nickel(111)single crystal studied by the integration of ultrafast reflectivity and polarimetry probes

引用

Chinese Physics B 2024年第3期33卷 65-69页

作者：匡皓余军潇陈洁 H.E.Elsayed-Ali 李润泽 Peter M.Rentzepis School of Physical Science and Technology ShanghaiTech UniversityShanghai 201210China Center for Transformative Science ShanghaiTech UniversityShanghai 201210China Center for Ultrafast Science and Technology Key Laboratory for Laser Plasmas (Ministry of Education)Collaborative Innovation Center of IFSA (CICIFSA)School of Physics and AstronomyShanghai Jiao Tong UniversityShanghai 200240China Department of Electrical and Computer Engineering Old Dominion UniversityNorfolkVirginia 23529USA Department of Electrical and Computer Engineering Texas A&M UniversityCollege StationTexas 77843USA

With the integration of ultrafast reflectivity and polarimetry probes,we observed carrier relaxation and spin dynamics induced by ultrafast laser excitation of Ni(111)single *** carrier relaxation time within the linear excitation range reveals that electron-phonon coupling and dissipation of photon energy into the bulk of the crystal take tens of *** the other hand,the observed spin dynamics indicate a longer time of about 120 *** further understand how the lattice degree of freedom is coupled with these dynamics may require the integration of an ultrafast diffraction probe.

关键词： ultrafast spin dynamics non-equilibrium dynamics multi-probe

来源：评论

学校读者我要写书评

暂无评论

Hybrid Recurrent Neural Network for Signal-Dependent Noise Suppression in Molecular Communication

引用

IEEE Transactions on Molecular, Biological, and Multi-Scale Communications 2025年第2期11卷 283-291页

作者： Xiang, Cheng Zhang, Yaqing Huang, Yu Tan, Weiqiang Chen, Xuan Wen, Miaowen Guangzhou University Research Center of Intelligent Communication Engineering School of Electronics and Communication Engineering Guangzhou510006 China Guangzhou University School of Computer Science and Cyber Engineering Guangzhou510006 China South China University of Technology Guangdong Provincial Key Laboratory of Short-Range Wireless Detection and Communication School of Electronic and Information Engineering Guangzhou510641 China

Molecular communication (MC) employs chemical molecules for information transfer in environments where electromagnetic signals are ineffective. However, the diffusion mechanism introduces signal-dependent noise (SDN), complicating accurate signal recovery. Traditional model-based methods struggle to handle SDN's complex dynamics and depend heavily on optimal parameter tuning, limiting their adaptability to temporal variations. To tackle these challenges, this paper introduces a hybrid recurrent neural network (RNN) model that effectively captures both short-and long-term dependencies within MC signals, surpassing the performance of single RNN models and traditional approaches. This model offers a promising data-driven solution for noise mitigation in MC, with its effectiveness validated through numerical simulation results. © 2015 IEEE.

关键词： Recurrent neural 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：