检索结果-内蒙古大学图书馆

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

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

Imp: Highly Capable Large Multimodal Models for Mobile Devices

引用

IEEE Transactions on Multimedia 2025年 27卷 2961-2974页

作者： Shao, Zhenwei Yu, Zhou Yu, Jun Ouyang, Xuecheng Zheng, Lihao Gai, Zhenbiao Wang, Mingyang Kuang, Zhenzhong Ding, Jiajun Hangzhou Dianzi University Key Laboratory of Complex Systems Modeling and Simulation School of Computer Science Hangzhou310005 China Harbin Institute of Technology Shenzhen Department of Computer Science and Technology Shenzhen China Hangzhou Dianzi University HDU-ITMO Joint Institute Hangzhou310005 China

By harnessing the capabilities of large language models (LLMs), recent large multimodal models (LMMs) have shown remarkable versatility in open-world multimodal understanding. Nevertheless, they are usually parameter-heavy and computation-intensive, thus hindering their applicability in resource-constrained scenarios. To this end, several lightweight LMMs have been proposed successively to maximize the capabilities under constrained scale (e.g., 3B). Despite the encouraging results achieved by these methods, most of them only focus on one or two aspects of the design space, and the key design choices that influence model capability have not yet been thoroughly investigated. In this paper, we conduct a systematic study for lightweight LMMs from the aspects of model architecture, training strategy, and training data. Based on our findings, we obtain Imp - a family of highly capable LMMs at the 2B$\sim$4B scales. Notably, our Imp-3B model steadily outperforms all the existing lightweight LMMs of similar size, and even surpasses the state-of-the-art LMMs at the 13B scale. With low-bit quantization and resolution reduction techniques, our Imp model can be deployed on a Qualcomm Snapdragon 8Gen3 mobile chip with a high inference speed of about 13 tokens/s. © 2025 IEEE.

关键词： Visual languages

来源：评论

学校读者我要写书评

暂无评论

Improved Mayfly Algorithm Based on Co-Evolution for Feature Selection

Improved Mayfly Algorithm Based on Co-Evolution for Feature ...

引用

2023 IEEE International Conference on Systems, Man, and Cybernetics, SMC 2023

作者： Tianyu, Guo Hongbin, Dong Jing, Zhou School of Computer Science and Technology Harbin Engineering University National Engineering Laboratory for E-government Modeling and Simulation Harbin China

ISBN: (纸本)9798350337020

Mayfly algorithm (MA) is a new type of intelligent optimization algorithm with excellent search ability and a broad application prospect in feature selection. However, as the dimension of the data increases, MA also has the problem of weak search ability and falling into the local optimal solution. To solve these problems, this paper combines the idea of collaborative evolution and proposes an improved Mayfly algorithm (CEIMA) suitable for feature selection problems. The CEIMA algorithm first divides the population into two sub-populations and initializes them in different ways to increase the diversity of the population. During evolution, information sharing is achieved through an information transmission mechanism between sub-populations, achieving collaborative evolution between sub-populations to enhance the global search ability of CEIMA. Finally, the mRMR-m strategy is proposed to mutate the global optimal individual based on feature importance, improving the ability of CEIMA to escape from local optimal solutions. Through a comparison with 7 feature selection algorithms on 12 UCI datasets, the effectiveness of CEIMA is proven. © 2023 IEEE.

关键词： co-evolution feature selection halton sequence max-relevance and min-redundancy mayfly algorithm

来源：评论

学校读者我要写书评

暂无评论

Potential Distribution of Seagrass Meadows Based on the MaxEnt Model in Chinese Coastal Waters

引用

Journal of Ocean University of China 2022年第5期21卷 1351-1361页

作者： WANG Ming WANG Yong LIU Guangliang CHEN Yuhu YU Naijing Center for High Performance Computing and System Simulation Pilot National Laboratory for Marine Science and TechnologyQingdao 266200China Shandong Provincial Key Laboratory of Computer Networks Qilu University of Technology(Shandong Academy of Sciences)Jinan 250101China

Seagrass meadows are generally diverse in China and have become important ecosystem with essential ***,the seagrass distribution across the seawaters of China has not been evaluated,and the magnitude and direction of changes in seagrass meadows remain *** study aimed to provide a nationwide seagrass distribution map and explore the dynamic changes in seagrass population under global climate *** studies were performed using the modeling software MaxEnt with 58961 occurrence records and 27 marine environmental variables to predict the potential distribution of seagrasses and calculate the *** environmental variables were excluded from the modeling processes based on a correlation analysis to ensure predicted *** predicted area was 790.09 km^(2),which is much larger than the known seagrass distribution in China(87.65 km^(2)).By 2100,the suitable habitat of seagrass will shift northwest and increase to 923.62 *** of the sum of the individual family under-pre-dicted the national distribution of seagrasses and consistently showed a downward trend in the *** of all environmental vari-ables,physical parameters(e.g.,depth,land distance,and sea surface temperature)contributed the most in predicting seagrass distri-butions,and nutrients(e.g.,nitrate,phosphate)ranked among the key influential predictors for habitat suitability in our study *** study is the first effort to fill a gap in understanding the distribution of seagrasses in *** studies using modeling and biological/ecological approaches are warranted.

关键词： seagrass meadows species distribution modeling global climate change Chinese coastal waters

来源：评论

学校读者我要写书评

暂无评论

Interference-assisted energy harvesting short packet communications with hardware impairments

引用

International Journal of Intelligent Networks 2024年第1期5卷 231-240页

作者： Chen, Dechuan Li, Jin Hu, Jianwei Zhang, Xingang Zhang, Shuai Wang, Dong College of Physics and Electronic Engineering Nanyang Normal University Nanyang473061 China Ministry of Education Nanjing210003 China National Key Laboratory for Complex Systems Simulation Beijing100101 China College of Computer Science and Technology Nanyang Normal University Nanyang473061 China

Radio frequency energy harvesting offers a promising solution to provide low power Internet of Things (IoT) devices with convenient and perpetual energy supply. In this work, we investigate the reliable performance of an energy-constrained transmitter communicating with a receiver over Nakagami-m channel, where the effects of transceiver hardware impairments and finite blocklength coding are jointly considered. Specifically, the communication link between the transmitter and receiver operates within the coverage of an existing wireless system, with radio frequency signal from the existing system serving as an energy signal for the transmitter while acting as an interference signal for the receiver. By utilizing the finite-blocklength information theory, we first derive average block error rate (BLER) and asymptotic average BLER in closed-form expressions, which enable us to quantify the extent of reliability loss. Then, we analyze effective throughput of the system, and determine the optimal blocklength that maximizes the effective throughput. computer simulations are employed to validate the accuracy of our analytical findings, demonstrating the presence of an outage threshold solely due to hardware impairments. Furthermore, if transmission rate exceeds the outage threshold defined by the level of hardware impairments, reliable communication within the system under consideration cannot be achieved, regardless of the transmit signal-to-noise ratio (SNR). © 2024 The Authors

关键词： Energy harvesting

来源：评论

学校读者我要写书评

暂无评论

An Anomaly Detection Method Based on Adaptive Log and Dual Feature Fusion Analysis for Distributed Systems 11

An Anomaly Detection Method Based on Adaptive Log and Dual F...

引用

11th International Conference on Information Systems and Computing technology, ISCTech 2023

作者： Jia, Yijiang Li, Kefei Lu, Ping Xie, Baodi Wang, Huayang Chen, Jincai Huazhong University of Science and Technology School of Computer Science and Technology Wuhan China Huazhong University of Science and Technology Wuhan National Laboratory for Optoelectronics Wuhan China National Key Laboratory of Modeling and Simulation for Complex Systems Beijing China

ISBN: (纸本)9798350342406

In this study, we present a middleware-based approach for detecting anomalies in distributed systems. Our method facilitates the dynamic collection of logs at various levels of detail and incorporates an a priori dictionary-based compression strategy to process and transmit logs, minimizing the performance impact on distributed systems. Additionally, we utilize a dual feature fusion technique to analyze the logs. We evaluate the effectiveness of our approach by performing anomaly detection in a publish/subscribe distributed system and comparing it with existing methods. The results illustrate that our method outperforms other approaches, demonstrating superior performance. © 2023 IEEE.

关键词： Anomaly detection

来源：评论

学校读者我要写书评

暂无评论

Detection of Detonators in Baggage Using Deep CNN Architecture: A Comparative Study of Contrast Enhancement and Feature Extraction Methods 8

Detection of Detonators in Baggage Using Deep CNN Architectu...

引用

8th IEEE International Conference on Image and Signal Processing and their Applications, ISPA 2024

作者： Lynda, Oulhissane Mostefa, Merah Moldovanu, Simona Moraru, Luminita LSS Faculty of Science and Technology Abdelhamid Ibn Badis University Laboratory of Signals and Systems Mostaganem Algeria Faculty of Automation Computers Electrical Engineering and Electronics Dunarea de Jos University of Galati Modelling & Simulation Laboratory MSlab Department of Computer Science and Information Technology Galati Romania Faculty of Sciences and Environment Modelling & Simulation Laboratory Dunarea de Jos University of Galati MSlab Department of Chemistry Physics & Environment Galati Romania

ISBN: (纸本)9798350309249

Visual inspection of dual-energy X-ray radiographic images of cabin baggage requires high performance, but is hindered by various challenges such as low target prevalence, variability in target visibility, possible presence of multiple targets, and security personnel fatigue and inattention. Artificial intelligence (AI) techniques, particularly deep convolutional neural networks (CNNs), have shown promise in improving the automatic detection of explosives, even with low-resolution radiographic images, especially in high baggage throughput scenarios. In this paper, we focus on the detection of detonators as components of improvised explosive devices. The proposed approach involves comparing two experiments implemented in a deep CNN architecture using TensorFlow and Keras libraries. In the first experiment, raw dual-energy radiographic images without any enhancement were used. The second experiment includes three methods for contrast enhancement and feature extraction: the Contrast Limited Adaptive Histogram Equalization (CLAHE) method, the wavelet transform-based method, and the mixed CLAHE RGB-Wavelet method. In the latter two methods, Haar, Db2, Coif2, and Sym2 mother wavelet functions at two levels (HH and HL) were employed. The analysis of results focuses on a comparative study of performance measures such as accuracy, precision, recall, and F1 score. It was found that the preprocessing methods used in experiment 2, for the two evaluated classes (detonator and no detonator), achieved higher accuracy compared to the raw radiographic images used in experiment 1 (98.08%). The highest accuracies in experiment 2, with a value of 100%, were obtained with the CLAHE method (green channel in grayscale, blue channel in grayscale, and RGB channels) and the wavelet transform method with Haar mother wavelet at two levels HL © 2024 IEEE.

关键词： Feature extraction

来源：评论

学校读者我要写书评

暂无评论

Evolutionary Interest Representation Network for Click-Through Rate Prediction

Evolutionary Interest Representation Network for Click-Throu...

引用

2023 IEEE International Conference on Web Services, ICWS 2023

作者： Zhou, Renjie Jin, Zhenming Wan, Jian Zhang, Jilin Guo, Xingyu Liu, Qiang Hangzhou Dianzi University Key Laboratory of Complex Systems Modeling and Simulation of the Ministry of Education School of Computer Science and Technology Hangzhou China Hangzhou Dianzi University Key Laboratory of Complex Systems Modeling and Simulation of the Ministry of Education Itmo Joint Institute Hangzhou China Zhejiang University of Science and Technology School of Information and Electronic Engineering Hangzhou China Zhejiang Police College Dept. of Computer and Information Technology Hangzhou China

ISBN: (纸本)9798350304855

The rapid development of the Internet has revo-lutionized our lives, providing us with an array of convenient services. However, this revolution has also led to the problem of data overload. Personalized recommendation services have been widely recognized as an effective solution to this problem by both the industrial and academic communities. Accurate prediction of click-through rate (CTR) is a critical task in personalized recommendation systems since it can improve the user's shopping experience, ultimately increasing revenue for the platform. To achieve accurate CTR prediction, it is crucial to capture user interests. While several methods for interest modeling exist, challenges related to user interest evolution and user behavior noise need further attention. This paper proposes a novel CTR prediction model named the Evolutionary Interest Representation Network (EIRN), which learns the evolving interest representation of users from their behaviors, profiles, and environmental attributes. In the model, we introduce an advertisement correlation graph decomposition layer and a noise filter layer to enhance and purify the user behavior representation. An interest extraction layer is designed to capture sequential correlations and common characteristics between user behaviors, representing their interests. Additionally, as user interests evolve over time and their importance changes, an attention-based interest evolution layer is designed to track the evolution of user interests. Our proposed method is validated using four public datasets and an industrial dataset, showing its superiority over existing CTR prediction methods. © 2023 IEEE.

关键词： Forecasting

来源：评论

学校读者我要写书评

暂无评论

Secure short-packet communications using a full-duplex receiver

引用

International Journal of Intelligent Networks 2023年第1期4卷 349-354页

作者： Chen, Dechuan Li, Jin Hu, Jianwei Zhang, Xingang Zhang, Shuai College of Physics and Electronic Engineering Nanyang Normal University Nanyang473061 China Ministry of Education Nanjing210003 China National Key Laboratory for Complex Systems Simulation Beijing100101 China College of Computer Science and Technology Nanyang Normal University Nanyang473061 China

In this work, we consider the physical layer security in short-packet communications, where a full-duplex (FD) receiver receives information signals from a source while generating artificial noise (AN) to confuse an eavesdropper. Taking into account the finite blocklength coding and the self-interference (SI) in FD mode, we derive new approximation closed-form expression for the secrecy throughput. Moreover, we analyze the asymptotic secrecy throughput in two distinct scenarios, i.e., high signal-to-noise ratio (SNR) regime and infinite blocklength coding, to gain more insights. Our examination illustrates the correctness of our expressions and shows how the critical system variables affect the secrecy throughput. © 2023 The Authors

关键词： Finite difference method

来源：评论

学校读者我要写书评

暂无评论

Two-Stage Point Cloud Super Resolution with Local Interpolation and Readjustment via Outer-Product Neural Network

引用

Journal of Systems Science & Complexity 2021年第1期34卷 68-82页

作者： WANG Guangyu XU Gang WU Qing WU Xundong School of Computer Science and Technology Hangzhou Dianzi UniversityHangzhou 310018China Key Laboratory of Complex Systems Modeling and Simulation Ministry of EducationHangzhou 310018China

This paper proposes a two-stage point cloud super resolution framework that combines local interpolation and deep neural network based readjustment. For the first stage, the authors apply a local interpolation method to increase the density and uniformity of the target point cloud. For the second stage, the authors employ an outer-product neural network to readjust the position of points that are inserted at the first stage. Comparison examples are given to demonstrate that the proposed framework achieves a better accuracy than existing state-of-art approaches, such as PU-Net, Point Net and DGCNN(Source code is available at https://***/qwerty1319/PC-SR).

关键词： Neural network outer-product network point cloud super resolution

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：