检索结果-内蒙古大学图书馆

Attar: RRAM-based in-memory attention accelerator with software-hardware co-optimization

Science China(information Sciences) 2025年第3期68卷 371-387页

作者： Bing LI Ying QI Ying WANG Yinhe HAN Information Engineering College Capital Normal University Research Center for Intelligent Computing Systems Institute of Computing TechnologyChinese Academy of Sciences

The attention mechanism has become a pivotal component in artificial intelligence, significantly enhancing the performance of deep learning applications. However, its quadratic computational complexity and intricate computations lead to substantial inefficiencies when processing long sequences. To address these challenges, we introduce Attar, a resistive random access memory(RRAM)-based in-memory accelerator designed to optimize attention mechanisms through software-hardware co-optimization. Attar leverages efficient Top-k pruning and quantization strategies to exploit the sparsity and redundancy of attention matrices, and incorporates an RRAM-based in-memory softmax engine by harnessing the versatility of the RRAM crossbar. Comprehensive evaluations demonstrate that Attar achieves a performance improvement of up to 4.88× and energy saving of 55.38% over previous computing-in-memory(CIM)-based accelerators across various models and datasets while maintaining comparable accuracy. This work underscores the potential of in-memory computing to enhance the efficiency of attention-based models without compromising their effectiveness.

关键词： RRAM computing-in-memory attention pruning quantization

来源：评论

学校读者我要写书评

暂无评论

SDGNN: Symmetry-Preserving Dual-Stream Graph Neural Networks

引用

IEEE/CAA Journal of Automatica Sinica 2024年第7期11卷 1717-1719页

作者： Jiufang Chen Ye Yuan Xin Luo the College of Computer Science and Technology Chongqing University of Posts and TelecommunicationsChongqing 400065 the Chongqing Key Laboratory of Big Data and Intelligent Computing Chongqing Institute of Green and Intelligent TechnologyChinese Academy of SciencesChongqing 400714China the College of Computer and Information Science Southwest UniversityChongqing 400715China IEEE

Dear Editor,This letter proposes a symmetry-preserving dual-stream graph neural network(SDGNN) for precise representation learning to an undirected weighted graph(UWG). Although existing graph neural networks(GNNs) are influential instruments for representation learning to a UWG, they invariably adopt a unique node feature matrix for illustrating the sole node set of a UWG.

关键词： representation preserving undirected

来源：评论

学校读者我要写书评

暂无评论

Audio-guided self-supervised learning for disentangled visual speech representations

引用

Frontiers of Computer Science 2024年第6期18卷 277-279页

作者： Dalu FENG Shuang YANG Shiguang SHAN Xilin CHEN Key Laboratory of Intelligent Information Processing of Chinese Academy of Sciences Institute of Computing TechnologyChinese Academy of SciencesBeijing 100190China University of Chinese Academy of Sciences Beijing 100049China

1 *** visual speech representations from talking face videos is an important problem for several speech-related tasks,such as lip reading,talking face generation,and audiovisual speech separation[1,2].The key difficulty lies in tackling speech-irrelevant factors presented in the videos,such as lighting,resolution,viewpoints,and head motion.

关键词： such dis visual

来源：评论

学校读者我要写书评

暂无评论

Leveraging Slot-Aware Enhancements for Improved Zero-Shot Dialogue State Tracking 6

Leveraging Slot-Aware Enhancements for Improved Zero-Shot Di...

引用

6th International Conference on Frontier Technologies of information and Computer, ICFTIC 2024

作者： Song, Tianhang Hu, Haolei Chen, Xinxin Zi, Kangli Henan Institute of Advanced Technology Zhengzhou University Zhengzhou China Institute of Computing Technology Key Laboratory of Intelligent Information Processing Beijing China

ISBN: (纸本)9798331541750

Task-oriented dialogue systems (TOD) aim to help users complete specific tasks through multiple rounds of dialogue, in which Dialogue State Tracking (DST) is a key component. The training of DST models typically necessitates the availability of a substantial corpus of dialogue data with turn-level annotations. However, the collection of such data is both costly and inefficient. Despite the impressive performance of systems based on advanced large language models (LLMs), their closed-source nature restricts the potential for local deployment. To address this challenge, this paper proposes a new zero-shot slot-aware enhanced model (SAE), which introduces a novel coded prompting framework to enhance the DST capability of small parameter models in new domains and task scenarios. In particular, we have devised a prompting framework that employs code to refine task instructions. Furthermore, we have proposed a slot-aware question generation mechanism that generates slot-aware questions based on contextual details in accordance with the dialogue context, thereby enhancing the model's perception and matching capabilities for slots. In order to verify the effectiveness of the model, we conducted experiments on the MultiWOZ dataset. The results show that the SAE model has achieved significant performance improvements in zero-shot DST tasks. © 2024 IEEE.

关键词： Zero-shot learning

来源：评论

学校读者我要写书评

暂无评论

From Static to Dynamic: Adapting Landmark-Aware Image Models for Facial Expression Recognition in Videos

引用

IEEE Transactions on Affective computing 2024年第2期16卷 624-638页

作者： Chen, Yin Li, Jia Shan, Shiguang Wang, Meng Hong, Richang Hefei University of Technology School of Computer Science and Information Engineering Hefei230601 China Institute of Computing Technology Chinese Academy of Sciences Key Laboratory of Intelligent Information Processing Beijing100190 China University of Chinese Academy of Sciences Beijing100049 China

Dynamic facial expression recognition (DFER) in the wild is still hindered by data limitations, e.g., insufficient quantity and diversity of pose, occlusion and illumination, as well as the inherent ambiguity of facial expressions. In contrast, static facial expression recognition (SFER) currently shows much higher performance and can benefit from more abundant high-quality training data. Moreover, the appearance features and dynamic dependencies of DFER remain largely unexplored. Recognizing the potential in leveraging SFER knowledge for DFER, we introduce a novel Static-to-Dynamic model (S2D) that leverages existing SFER knowledge and dynamic information implicitly encoded in extracted facial landmark-aware features, thereby significantly improving DFER performance. Firstly, we build and train an image model for SFER, which incorporates a standard Vision Transformer (ViT) and Multi-View Complementary Prompters (MCPs) only. Then, we obtain our video model (i.e., S2D), for DFER, by inserting Temporal-Modeling Adapters (TMAs) into the image model. MCPs enhance facial expression features with landmark-aware features inferred by an off-the-shelf facial landmark detector. And the TMAs capture and model the relationships of dynamic changes in facial expressions, effectively extending the pre-trained image model for videos. Notably, MCPs and TMAs only increase a fraction of trainable parameters (less than +10%) to the original image model. Moreover, we present a novel Emotion-Anchors (i.e., reference samples for each emotion category) based Self-Distillation Loss to reduce the detrimental influence of ambiguous emotion labels, further enhancing our S2D. Experiments conducted on popular SFER and DFER datasets show that we have achieved a new state of the art. © 2010-2012 IEEE.

关键词： Face recognition

来源：评论

学校读者我要写书评

暂无评论

Non-uniform Compressive Sensing Imaging Based on Image Saliency

引用

Chinese Journal of Electronics 2023年第1期32卷 159-165页

作者： LI Hongliang DAI Feng ZHAO Qiang MA Yike CAO Juan ZHANG Yongdong University of Chinese Academy of Sciences Key Lab of Intelligent Information Processing of Chinese Academy of Sciences Institute of Computing Technology Chinese Academy of Sciences University of Science and Technology of China

For more effective image sampling, compressive sensing(CS) imaging methods based on image saliency have been proposed in recent years. Those methods assign higher measurement rates to salient regions,but lower measurement rate to non-salient regions to improve the performance of CS imaging. However, those methods are block-based, which are difficult to apply to actual CS sampling, as each photodiode should strictly correspond to a block of the scene. In our work, we propose a non-uniform CS imaging method based on image saliency, which assigns higher measurement density to salient regions and lower density to non-salient regions,where measurement density is the number of pixels measured in a unit size. As the dimension of the signal is reduced, the quality of reconstructed image will be improved theoretically, which is confirmed by our experiments. Since the scene is sampled as a whole, our method can be easily applied to actual CS sampling. To verify the feasibility of our approach, we design and implement a hardware sampling system, which can apply our non-uniform sampling method to obtain measurements and reconstruct the images. To our best knowledge, this is the first CS hardware sampling system based on image saliency.

关键词： Measurement units Density measurement Imaging Image sampling Size measurement Hardware Sensors

来源：评论

学校读者我要写书评

暂无评论

A Multi-Hyperparameter Prediction Framework for Distributed Energy Trading on Photovoltaic Network

引用

Tsinghua Science and technology 2025年第2期30卷 864-874页

作者： Chun Chen Yong Zhang Boon Han Lim Li Ning Shengzhong Feng Peng Xie School of Intelligent Manufacturing and Equipment Shenzhen Institute of Information TechnologyShenzhen 518172China Shenzhen Institute of Advanced Technology Chinese Academy of SciencesShenzhen518055China Lee Kong Chian Faculty of Engineering and Science Universiti Tunku Abdul RahmanKajang 43000Malaysia Shenzhen Institute for Advanced Study Shenzhen 518028China Shenzhen Institute of Advanced Technology Chinese Academy of SciencesBeijing 100864China High-performance Intelligent Computing Research Group Guangdong Institute of Intelligent Science and TechnologyZhuhai 519031China

The rapid evolution of distributed energy resources,particularly photovoltaic systems,poses a formidable challenge in maintaining a delicate balance between energy supply and demand while minimizing *** integrated nature of distributed markets,blending centralized and decentralized elements,holds the promise of maximizing social welfare and significantly reducing overall costs,including computational and communication ***,achieving this balance requires careful consideration of various hyperparameter sets,encompassing factors such as the number of communities,community detection methods,and trading mechanisms employed among *** address this challenge,we introduce a groundbreaking neural network-based framework,the Energy Trading-based Artificial Neural Network(ET-ANN),which excels in performance compared to existing *** experiments underscore the superiority of ET-ANN in minimizing total energy transaction costs while maximizing social welfare within the realm of photovoltaic networks.

关键词： photovoltaic energy trading hyperparameter prediction

来源：评论

学校读者我要写书评

暂无评论

PFedKD: Personalized Federated Learning via Knowledge Distillation Using Unlabeled Pseudo Data for Internet of Things

引用

IEEE Internet of Things Journal 2025年第11期12卷 16314-16324页

作者： Li, Hanxi Chen, Guorong Wang, Bin Chen, Zheng Zhu, Yongsheng Hu, Fuqiang Dai, Jiao Wang, Wei Beijing Jiaotong University Beijing Key Laboratory of Security and Privacy in Intelligent Transportation Beijing100044 China Institute of Computing Technologies China Railway Information Technology Group Corporation Ltd. Beijing10081 China

With the rapid advancement of wearable devices and Internet of Things (IoT) technologies, sensor data generated by edge devices has surged. This data is crucial for advancing IoT applications, including health status monitoring, abnormal behavior detection, and environmental monitoring. However, traditional centralized learning requires uploading data to a central server, raising security and privacy concerns and hindering data application. Federated learning (FL) offers a solution by enabling collaborative model training on IoT devices without transferring data from the local device. In practice, edge devices generate data that is often highly heterogeneous, making it challenging for the global FL model to capture local data distributions accurately, leading to significant performance degradation. Additionally, imbalanced edge device resources and limited bandwidth can cause data transmission delays or interruptions, impacting application feasibility. To address these issues, we propose PFedKD, a novel personalized FL algorithm based on knowledge distillation, aimed at enhancing the model’s generalization ability and reducing communication overhead in heterogeneous IoT data environments. PFedKD constructs a public dataset using unlabeled pseudo data to extract knowledge from each client, training personalized models that fit local data distributions. This method controls dataset size while enhancing performance. During communication, only logits and class prototypes are transmitted, ensuring high communication efficiency. Sharpness aware minimization is introduced in local model training to optimize generalization. Additionally, we design a weight distribution mechanism based on client sample quality evaluation that optimizes knowledge aggregation and model personalization. Extensive experiments demonstrate that PFedKD significantly outperforms state-of-the-art baselines in both learning performance and communication efficiency. © 2014 IEEE.

关键词： Wearable technology

来源：评论

学校读者我要写书评

暂无评论

DMSS:An Attention-Based Deep Learning Model for High-Quality Mass Spectrometry Prediction

引用

Big Data Mining and Analytics 2024年第3期7卷 577-589页

作者： Yihui Ren Yu Wang Wenkai Han Yikang Huang Xiaoyang Hou Chunming Zhang Dongbo Bu Xin Gao Shiwei Sun Key Lab of Intelligent Information Processing Institute of Computing TechnologyChinese Academy of SciencesBeijing 100190Chinaand with University of Chinese Academy of SciencesBeijing 100049China Syneron Technology Guangzhou 510000China College of Information and Electrical Engineering China Agricultural UniversityBeijing 100083China Computer Science Program ComputerElectrical and Mathematical Sciences and Engineering DivisionKing Abdullah University of Science and Technology(KAUST)Thuwal 23955-6900Kingdom of Saudi Arabia Insitute of Computing Technology Chinese Academy of SciencesBeijing 100190China with Western Institute of Computing Technology Chongqing 400000China

Accurate prediction of peptide spectra is crucial for improving the efficiency and reliability of proteomic analysis,as well as for gaining insight into various biological *** this study,we introduce Deep MS Simulator(DMSS),a novel attention-based model tailored for forecasting theoretical spectra in mass *** has undergone rigorous validation through a series of experiments,consistently demonstrating superior performance compared to current methods in forecasting theoretical *** superior ability of DMSS to distinguish extremely similar peptides highlights the potential application of incorporating our predicted intensity information into mass spectrometry search engines to enhance the accuracy of protein *** findings contribute to the advancement of proteomics analysis and highlight the potential of the DMSS as a valuable tool in the field.

关键词： mass spectrometry proteomics machine learning deep learning

来源：评论

学校读者我要写书评

暂无评论

Policy Network-Based Dual-Agent Deep Reinforcement Learning for Multi-Resource Task Offloading in Multi-Access Edge Cloud Networks

引用

China Communications 2024年第4期21卷 53-73页

作者： Feng Chuan Zhang Xu Han Pengchao Ma Tianchun Gong Xiaoxue School of Communications and Information Engineering Chongqing University of Posts and TelecommunicationsChongqing 400065China Institute of Intelligent Communication and Network Security Chongqing University of Posts and TelecommunicationsChongqing 400065China Key Laboratory of Big Data Intelligent Computing Chongqing University of Posts and TelecommunicationsChongqing 400065China School of Information Engineering Guangdong University of TechnologyGuangzhou 510006China

The Multi-access Edge Cloud(MEC) networks extend cloud computing services and capabilities to the edge of the networks. By bringing computation and storage capabilities closer to end-users and connected devices, MEC networks can support a wide range of applications. MEC networks can also leverage various types of resources, including computation resources, network resources, radio resources,and location-based resources, to provide multidimensional resources for intelligent applications in 5/***, tasks generated by users often consist of multiple subtasks that require different types of resources. It is a challenging problem to offload multiresource task requests to the edge cloud aiming at maximizing benefits due to the heterogeneity of resources provided by devices. To address this issue,we mathematically model the task requests with multiple subtasks. Then, the problem of task offloading of multi-resource task requests is proved to be NP-hard. Furthermore, we propose a novel Dual-Agent Deep Reinforcement Learning algorithm with Node First and Link features(NF_L_DA_DRL) based on the policy network, to optimize the benefits generated by offloading multi-resource task requests in MEC networks. Finally, simulation results show that the proposed algorithm can effectively improve the benefit of task offloading with higher resource utilization compared with baseline algorithms.

关键词： benefit maximization deep reinforcement learning multi-access edge cloud task offloading

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：