检索结果-内蒙古大学图书馆

International Conference on Data Science and Advanced Analytics (DSAA)

作者： Junfeng Wu Guangyan Huang Hui Zheng Guang-Li Huang Yu Hu Jing He School of Information Technology Deakin University Burwood Australia State Key Laboratory of Computer Architecture University of Chinese Acamdemy of Sciences Beijing China University of Oxford Oxford United Kingdom

ISBN: (纸本)9781665473316

The success of advanced Background Subtraction (BGS) algorithms for dynamic backgrounds is mostly in land scenes such as those in CDNet benchmarks; few handle underwater scenes, since existing underwater video datasets are either in low resolution or with only static backgrounds. Consequently, the lack of reliable BGS support makes supervised Moving-Objects Segmentation (MOS) algorithms much harder to adapt to unknown underwater scenes because of the diversities of the aquatic environments. For example, those trained by the latest underwater image dataset, SUIM, are ineffective in the underwater videos of our *** underwater waving objects (e.g., plants) often render existing BGS algorithms inaccurate due to three types of errors: (a) incompletely identified MOs (Moving Objects), (b) missing MOs, and (c) falsely identified MOs. In this paper, we propose a novel Clustering-Based Multi-state Background Representation (CBMSBR) model to learn and represent the repeatable patterns of waving movements in k background states (i.e., color ranges) per pixel, and thus accurately subtract the background waving objects to reduce these errors. In addition, we further develop a CBMSBR+ model to remove the more challenging background objects in unusually large magnitudes of wavings. Both models come from a basic observation: the video pixels in the waving zones repeatedly switch among multiple background states; e.g., a pixel switches among water state, plant 1 state, and plant 2 state. To test our proposed models, we create experiments using three types of challenging scenarios that each often covers at least two error types, i.e., the scattered MOs scenario covering (b) and (c), the crowded MOs scenario covering (a) - (c), and the slow MOs scenario covering (a) and (c). Experiments on these scenarios demonstrate the accuracy, effectiveness, and efficiency of our models and their applications in MOS improvements.

关键词： Image segmentation Image resolution Image color analysis Heuristic algorithms Switches Data science Object recognition

来源：评论

学校读者我要写书评

暂无评论

DU-DARTS: Decreasing the Uncertainty of Differentiable architecture Search 32

DU-DARTS: Decreasing the Uncertainty of Differentiable Archi...

引用

32nd British Machine Vision Conference, BMVC 2021

作者： Lu, Shun Hu, Yu Yang, Longxing Sun, Zihao Mei, Jilin Zeng, Yiming Li, Xiaowei Research Center for Intelligent Computing Systems Institute of Computing Technology Chinese Academy of Sciences China State Key Laboratory of Computer Architecture Institute of Computing Technology Chinese Academy of Sciences China School of Computer Science and Technology University of Chinese Academy of Sciences China Tecent ADlab China

Differentiable Neural architecture Search (DARTS) recently attracts a lot of research attention because of its high efficiency. However, the competition of candidate operations in DARTS introduces high uncertainty for selecting the truly important operation, thus leading to serious performance collapse. In this work, we decrease the uncertainty of differentiable architecture search (DU-DARTS) by enforcing the distribution of architecture parameters to approach the one-hot categorical distribution and by replacing the zero operation with a gate switch. Without any extra search cost, our method achieves state-of-the-art performance with 2.32%, 16.74%, and 24.1% test error on CIFAR-10, CIFAR-100, and ImageNet datasets, respectively. Moreover, DU-DARTS can robustly find an excellent architecture on NAS-Bench-1Shot1 and NAS-Bench-201, which further demonstrates the effectiveness of our method. The source code is available at https://***/ShunLu91/DU-DARTS. © 2021. The copyright of this document resides with its authors.

关键词： computer vision

来源：评论

学校读者我要写书评

暂无评论

Accelerating DNN-based 3D point cloud processing for mobile computing

引用

Science China(Information Sciences) 2019年第11期62卷 40-50页

作者： Bosheng LIU Xiaoming CHEN Yinhe HAN Jiajun LI Haobo XU Xiaowei LI State Key Laboratory of Computer Architecture Institute of Computing TechnologyChinese Academy of Sciences University of Chinese Academy of Sciences

3D point cloud data, which are produced by various 3 D sensors such as LIDAR and stereo cameras, have been widely deployed by industry leaders such as Google, Uber, Tesla, and Mobileye, for mobile robotic applications such as autonomous driving and humanoid robots. Point cloud data, which are composed of reliable depth information, can provide accurate location and shape characteristics for scene understanding, such as object recognition and semantic segmentation. However, deep neural networks(DNNs), which directly consume point cloud data, are particularly computation-intensive because they have to not only perform multiplication-and-accumulation(MAC) operations but also search neighbors from the irregular 3 D point cloud data. Such a task goes beyond the capabilities of general-purpose processors in realtime to figure out the solution as the scales of both point cloud data and DNNs increase from application to application. We present the first accelerator architecture that dynamically configures the hardware onthe-fly to match the computation of both neighbor point search and MAC computation for point-based DNNs. To facilitate the process of neighbor point search and reduce the computation costs, a grid-based algorithm is introduced to search neighbor points from a local region of grids. Evaluation results based on the scene recognition and segmentation tasks show that the proposed design harvests 16.4× higher performance and saves 99.95% of energy than an NVIDIA Tesla K40 GPU baseline in point cloud scene understanding applications.

关键词： deep neural network acceleration point cloud data neighbor point search mobile robotics hardware architecture

来源：评论

学校读者我要写书评

暂无评论

A Direction Vector-Guided Multi-Objective Evolutionary Algorithm for Variable Linkages Problems

SSRN

引用

SSRN 2023年

作者： Gu, Qinghua Zhang, Shaopeng Wang, Qian Xiong, Neal N. School of Resources Engineering Xi’an University of Architecture and Technology Shaanxi Xi’an710055 China Xi’an Key Laboratory for Intelligent Industrial Perception Calculation and Decision Xi'an University of Architecture and Technology Shaanxi Xi’an710055 China Department of Mathematics and Computer Science Northeastern State University TahlequahOK United States School of Management Xi’an University of Architecture and Technology Shaanxi Xi’an710055 China

Recombination is a key component for the evolutionary algorithm to provide promising offspring solutions. However, conventional recombination operators cannot generate high-quality solutions for variable linkages problems due to the particularity of Pareto sets(PS). To tackle this problem, a two-stage multi-objective evolutionary algorithm based on direction vector guidance(DSMOEA) is proposed in this paper. Firstly, a portion of the population is transformed by a covariance matrix to increase the probability of generating high-quality offspring. Then the representative solutions are selected in the transformed population to create the direction vectors. Under the guidance of the direction vectors, the population rapidly approaches PS and generates promising offspring solutions. Finally, DE is performed for searching globally to increase the diversity of the population. The proposed algorithm is tested on RM, IM, and UF test problems with 30, 50, and 100 dimensions to verify its performance. The experimental results show that the algorithm is promising for variable linkages problems. © 2023, The Authors. All rights reserved.

关键词： Covariance matrix

来源：评论

学校读者我要写书评

暂无评论

Aware: Adaptive Distributed Training with Computation, Communication and Position Awareness for Deep Learning Model

Aware: Adaptive Distributed Training with Computation, Commu...

引用

IEEE International Conference on High Performance Computing and Communications (HPCC)

作者： Yan Zeng Guangzheng Yi Yuyu Yin Jiyang Wu Meiting Xue Jilin Zhang Jian Wan Yunquan Zhang School of Computer Science and Technology Hangzhou Dianzi University Hangzhou China School of Cyberspace Hangzhou Dianzi University Hangzhou China State Key Laboratory of Computer Architecture Institute of Computing Technology Chinese Academy of Sciences Beijing China

The accuracy of the neural networks can usually be improved by increasing the size of the dataset and the layers or operators of the network, as it has strong composability. But, it makes a challenge to train these models efficiently, due to the limited resources of a single device, such as memory, computational resource and so on. For this challenge, it has become an inevitable trend to automatically implement model parallelism across multiple devices. This paper proposes Aware, an adaptive distributed parallel training method, to search distributed parallel strategy automatically for deep learning model. It firstly proposes an operator fusion strategy with computation and communication awareness to simplify the computational graph of deep learning model. And then, it introduces position-aware graph embedding algorithm to extract the structural features of models, which can make the searched parallel strategies to transplant to other networks with similar structures. On this basis, we use reinforcement learning algorithm to search distributed parallel strategy automatically. This paper makes experiments with neural networks such as Inception V3, NMT, GNM, NasNet and ResNet, and uses CIFAR10 and PTB datasets to compare the two methods of Hierarchical and Placeto. The results show that compared with Placeto, Aware achieves up to 5% reductions in runtime. Aware achieves more than 8% reductions in runtime compared with Hierarchical. Moreover, it has better transplantation and generalization capabilities, and supports pre-training for large-scale network parallel strategy search and accelerates convergence.

关键词： Deep learning Training Adaptation models Runtime Computational modeling Neural networks Reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Poster: Revisiting linpack algorithm on large-scale CPU-GPU heterogeneous systems 25

Poster: Revisiting linpack algorithm on large-scale CPU-GPU ...

引用

25th ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming, PPoPP 2020

作者： Shui, Chaoyang Yu, Xianzhi Yan, Yujin Wang, Yinshan Meng, Ke Tan, Guangming State Key Laboratory of Computer Architecture Institute of Computing Technology Chinese Academy of Sciences China University of Chinese Academy of Sciences China

ISBN: (纸本)9781450368186

As the widening gap between GPU computing capability and other components (CPU, PCIe bus and communication network), it’s increasingly challenging to design high performance parallel algorithms for large CPU-GPU heterogeneous systems. There are mainly two reasons. Firstly, simply offloading the kernel library to GPU incurs large volume data transfer through low-speed PCIe bus. Secondly, communication overheads through network severely affects scalability. To solve the above issues, we advocate a paradigm shift to GPU-centric and fine-grained pipelining algorithm design. By taking Linpack benchmark as a case study, the new algorithm design paradigm shows its effectiveness. Our optimized Linpack program achieves 63.79PFlops on 16384 GPUs. Its floating-point efficiency outperforms the NVIDIA proprietary counterparts by 5% on average. © 2020 Copyright held by the owner/author(s).

关键词： Graphics processing unit

来源：评论

学校读者我要写书评

暂无评论

An Auto-Parallel Method for Deep Learning Models Based on Genetic Algorithm

An Auto-Parallel Method for Deep Learning Models Based on Ge...

引用

International Conference on Parallel and Distributed Systems (ICPADS)

作者： Yan Zeng ChengChuang Huang YiJie Ni ChunBao Zhou JiLin Zhang Jue Wang MingYao Zhou MeiTing Xue YunQuan Zhang School of Computer Science and Technology Hangzhou Dianzi University Hangzhou China Key Laboratory for Modeling and Simulation of Complex Systems Ministry of Education Hangzhou China Data Security Governance Zhejiang Engineering Research Center Hangzhou China School of ITMO Joint Institute Hangzhou Dianzi University Hangzhou China Institute of Computer Network Information Center of the Chinese Academy of Sciences Beijing China HuaWei State Key Laboratory of Computer Architecture Institute of Computing Technology of the Chinese Academy of Sciences Beijing China

As the size of datasets and neural network models increases, automatic parallelization methods for models have become a research hotspot in recent years. The existing auto-parallel methods based on machine learning or graph algorithms still have issues with search efficiency and applicability. This paper proposes an automatic parallel method based on a dual-population genetic algorithm, TGA, which transforms model partitioning and placement into an integer linear programming problem and constructs a cost model to evaluate the solution. The solution space is built using the neural network’s dataflow graph and device cluster’s topology, and the dual-population genetic algorithm is used to search for the optimal model parallel strategy. Experiments with various models show that the proposed method can improve single-step execution time by up to 42% compared to the Baechi method and up to 37.7% compared to the Hierarchical method.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Alleviating Imbalance in Synchronous Distributed Training of Deep Neural Networks

Alleviating Imbalance in Synchronous Distributed Training of...

引用

IEEE International Conference on Big Data and Cloud Computing (BdCloud)

作者： Haiyang Lin Mingyu Yan Duo Wang Wenming Li Xiaochun Ye Zhimin Tang Dongrui Fan State Key Laboratory of Computer Architecture ICT CAS Beijing China University of Chinese Academy of Sciences Beijing China State Key Laboratory of Mathematical Engineering and Advanced Computing Wuxi China

Deep neural networks achieve huge success in computer vision, natural language process and other fields. As larger networks and larger datasets bring better performance, the training time grows rapidly. Distributed training uses parallelization strategy to accelerate the training of deep neural network. In the existing distributed training framework of deep neural network, the parameter server architecture has poor scalability and is not suitable for large-scale distributed training. Ring/HD All Reduce algorithm perform well in scalability under ideal circumstances. However, in many realistic scenarios, imbalance will occur frequently in the distributed training system, which is mainly the imbalance of compute ability and network. Due to the structure characteristics of synchronous distributed training and All Reduce algorithm, imbalance will bring serious impact on performance. At present, there is a lack of research on both compute ability imbalance and network imbalance, especially dynamic imbalance. To alleviate this problem, we analyze All Reduce algorithm first, and propose RingDL, which optimizes the performance of distributed training system by alleviating imbalance through customizable dynamic workloads strategy and link strategy. The experimental results show that RingDL significantly reduces idle time in computing phase and synchronous phase, improving the overall performance.

关键词： Training Deep learning Degradation computer vision Heuristic algorithms Scalability Neural networks

来源：评论

学校读者我要写书评

暂无评论

An Improved Weighted Ensemble Clustering Based on Two-Tier Uncertainty Measurement

SSRN

引用

SSRN 2022年

作者： Gu, Qinghua Wang, Peipei Li, Xuexian Chen, Lu Xiong, Neal N. School of Management Xi’an University of Architecture and Technology Xi’an China Xi’ an Key Laboratory for Intelligent Industrial Perception Calculation and Decision Xi'an University of Architecture and Technology Xi’an China Department of Mathematics and Computer Science Northeastern State University OK United States

Existing local weighted ensemble clustering algorithms strive to weight on each cluster, which take into account the differences among all clusters. However, the base clustering label on each cluster is ignored. The purpose of this paper is to combine the influence of cluster level and base clustering level in a unified ensemble clustering framework. A novel two-level weighted ensemble cluster method (TWEC) is proposed, which inserts global weighting strategy into a locally ensemble cluster learning framework. First, the cluster uncertainty based an entropy criterion is refined by considering the base clustering labels for each cluster. Then, two-level uncertainty is converted to cluster reliability via improved ensemble-driven cluster validity measure. Finally, two novel consensus functions are developed. The effectiveness of the proposed framework TWEC is verified against fourteen real-world datasets and ten comparison algorithms. The results demonstrate that the TWEC framework can improve the robustness and stability of the clustering. © 2022, The Authors. All rights reserved.

关键词： Clustering algorithms

来源：评论

学校读者我要写书评

暂无评论

Streamline Ring ORAM Accesses through Spatial and Temporal Optimization

Streamline Ring ORAM Accesses through Spatial and Temporal O...

引用

IEEE Symposium on High-Performance computer architecture

作者： Dingyuan Cao Mingzhe Zhang Hang Lu Xiaochun Ye Dongrui Fan Yuezhi Che Rujia Wang Tsinghua University Beijing China State Key Laboratory of Computer Architecture ICT CAS Beijing China State Key Laboratory of Mathematical Engineering and Advanced Computing China Illinois Institute of Technology

Memory access patterns could leak temporal and spatial information in a sensitive program; therefore, obfuscated memory access patterns are desired from the security perspective. Oblivious RAM (ORAM) has been the favored candidate to eliminate the access pattern leakage through randomly remapping data blocks around the physical memory space. Meanwhile, accessing memory with ORAM protocols results in significant memory bandwidth overhead. For each memory request, after going through the ORAM obfuscation, the main memory needs to service tens of actual memory accesses, and only one real access out of them is useful for the program execution. Besides, to ensure the memory bus access patterns are indistinguishable, extra dummy blocks need to be stored and transmitted, which cause memory space waste and poor performance. In this work, we introduce a new framework, String ORAM, that accelerates the Ring ORAM accesses with Spatial and Temporal optimization schemes. First, we identify that dummy blocks could significantly waste memory space and propose a compact ORAM organization that leverages the real blocks in memory to obfuscate the memory access pattern. Then, we identify the inefficiency of current transaction-based Ring ORAM scheduling on DRAM devices and propose an effective scheduling technique that can overlap the time spent on row buffer misses while ensuring correctness and security. With a minimal modification on the hardware and software, and negligible impact on security, the framework reduces 30.05% execution time and up to 40% memory space overhead compared to the state-of-the-art bandwidth-efficient Ring ORAM.

关键词： Protocols Space debris Memory management Random access memory Performance gain Scheduling Software

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：