检索结果-内蒙古大学图书馆

CMD: A Cache-assisted GPU Memory Deduplication Architecture

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Zhao, Wei Feng, Dan Tong, Wei Wei, Xueliang Wu, Bing The Wuhan National Laboratory for Optoelectronics Key Laboratory of Information Storage System Engineering Research Center of Data Storage Systems and Technology Ministry of Education of China School of Computer Science and Technology Huazhong University of Science and Technology China

Massive off-chip accesses in GPUs are the main performance bottleneck, and we divided these accesses into three types: (1) Write, (2) data-Read, and (3) Read-Only. Besides, We find that many writes are duplicate, and the duplication can be inter-dup and intra-dup. While inter-dup means different memory blocks are identical, and intra-dup means all the 4B elements in a line are the same. In this work, we propose a cache-assisted GPU memory deduplication architecture named CMD to reduce the off-chip accesses via utilizing the data duplication in GPU applications. CMD includes three key design contributions which aim to reduce the three kinds of accesses: (1) A novel GPU memory deduplication architecture that removes the inter-dup and inter-dup lines. As for the inter-dup detection, we reduce the extra read requests caused by the traditional read-verify hash process. Besides, we design several techniques to manage duplicate blocks. (2) We propose a cache-assisted read scheme to reduce the reads to duplicate data. When an L2 cache miss wants to read the duplicate block, if the reference block has been fetched to L2 and it is clean, we can copy it to the L2 missed block without accessing off-chip DRAM. As for the reads to intra-dup data, CMD uses the on-chip metadata cache to get the data. (3) When a cache line is evicted, the clean sectors in the line are invalidated while the dirty sectors are written back. However, most read-only victims are re-referenced from DRAM more than twice. Therefore, we add a full-associate FIFO to accommodate the read-only (it is also clean) victims to reduce the re-reference counts. Experiments show that CMD can decrease the off-chip accesses by 31.01%, reduce the energy by 32.78% and improve performance by 37.79%. Besides, CMD can improve the performance of memory-intensive workloads by 50.18%. © 2024, CC BY-NC-ND.

关键词： Cache memory

Generating Targeted Universal Adversarial Perturbation against Automatic Speech Recognition via Phoneme Tailoring

学校读者我要写书评

暂无评论

Generating Targeted Universal Adversarial Perturbation again...

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Yujun Zhang Yanqu Chen Jiakai Wang Jin Hu Renshuai Tao Xianglong Liu State Key Laboratory of Complex & Critical Software Environment Beihang University China School of Computer Science and Engineering Beihang University China College of Computer Science Beijing University of Technology China Zhongguancun Laboratory China School of Computer and Information Technology Beijing Jiaotong University China Institute of Data Space Hefei Comprehensive National Science Center China

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

There is a growing concern about adversarial attacks against automatic speech recognition (ASR) systems. Although research into targeted universal adversarial examples (AEs) has progressed, current methods are constrained by inefficient exploitation of audio features, demonstrating insufficient attack ability and robustness in the physical world. To solve this problem, we propose a phoneme-tailored attack (PTA) to improve the quality of the generated AEs. Specifically, to improve attack ability, we propose a Diverse Audio Composition Enrichment method, which enhances the utilization of audio features through phoneme-level slicing and recombination. To adapt AEs to complex environments, we propose a Natural Noise Pattern Guidance method to align AEs with natural noise patterns to improve their robustness. Experiments show that our method achieves an average accuracy of more than 72.34% and 98% with and without a norm constraint, and also demonstrates excellent performance in terms of generalization across datasets and resilience to MP3 compression.

关键词： Accuracy Target recognition Perturbation methods Noise Robustness Acoustics Speech processing Digital audio players Automatic speech recognition Resilience

SP3: Enhancing Structured Pruning via PCA Projection

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Hu, Yuxuan Zhang, Jing Zhao, Zhe Zhao, Chen Chen, Xiaodong Li, Cuiping Chen, Hong School of Information Renmin University of China Beijing China Key Laboratory of Data Engineering and Knowledge Engineering MOE China Engineering Research Center of Database and Business Intelligence MOE China Tencent AI Lab Tencent Beijing China School of Computer Science and Technology Xi'an Jiaotong University Xi'An China

Structured pruning is a widely used technique for reducing the size of pre-trained language models (PLMs), but current methods often overlook the potential of compressing the hidden dimension (d) in PLMs, a dimension critical to model size and efficiency. This paper introduces a novel structured pruning approach, Structured Pruning with PCA Projection (SP3), targeting the effective reduction of d by projecting features into a space defined by principal components before masking. Extensive experiments on benchmarks (GLUE and SQuAD) show that SP3 can reduce d by 70%, compress 94% of the BERTbase model, and maintain over 96% accuracy and outperform other methods that compress d by 6% in accuracy at the same compression ratio. SP3 has also proven effective with other models, including OPT and Llama. Our data and code are available at ours repo. © 2023, CC BY.

关键词：

Multi-Representation Genetic Programming: A Case Study on Tree-based and Linear Representations

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Huang, Zhixing Mei, Yi Zhang, Fangfang Zhang, Mengjie Banzhaf, Wolfgang Centre for Data Science and Artificial Intelligence School of Engineering and Computer Science Victoria University of Wellington Wellington6140 New Zealand Department of Computer Science and Engineering BEACON Center for the Study of Evolution in Action and Ecology Evolution and Behavior Program Michigan State University East LansingMI48864 United States

Existing genetic programming (GP) methods are typically designed based on a certain representation, such as tree-based or linear representations. These representations show various pros and cons in different domains. However, due to the complicated relationships among representation and fitness landscapes of GP, it is hard to intuitively determine which GP representation is the most suitable for solving a certain problem. Evolving programs (or models) with multiple representations simultaneously can alternatively search on different fitness landscapes since representations are highly related to the search space that essentially defines the fitness landscape. Fully using the latent synergies among different GP individual representations might be helpful for GP to search for better solutions. However, existing GP literature rarely investigates the simultaneous effective use of evolving multiple representations. To fill this gap, this paper proposes a multi-representation GP algorithm based on tree-based and linear representations, which are two commonly used GP representations. In addition, we develop a new cross-representation crossover operator to harness the interplay between tree-based and linear representations. Empirical results show that navigating the learned knowledge between basic tree-based and linear representations successfully improves the effectiveness of GP with solely tree-based or linear representation in solving symbolic regression and dynamic job shop scheduling problems. © 2024, CC BY.

关键词： Genetic programming

EABC: Energy-aware Centrality-based Caching for Named data Networking in the IoT

学校读者我要写书评

暂无评论

EABC: Energy-aware Centrality-based Caching for Named Data N...

IEEE International Symposium on World of Wireless Mobile and Multimedia Networks (WoWMoM)

作者： Xingyun He Hu Liu Wuyungerile Li Alvin Valera Winston K.G. Seah Inner Mongolia Key Lab. of Wireless Networking & Mobile Computing Engineering Research Center of Ecological Big Data Ministry of Education School of Computer Science Inner Mongolia University Hohhot China Wireless Networks Research Group School of Engineering and Computer Science Victoria University of Wellington Wellington New Zealand

ISBN: (数字)9798350394665

ISBN: (纸本)9798350394672

Named data Networking (NDN) is an information-centric internet architecture that delivers packets based on the name of the content in the packet. A key component of NDN is the caching strategy designed to reduce total network latency and load on content producers. To improve the speed and reliability of web content delivery, existing caching strategies typically cache content on a large number of intermediate nodes, which incur significant energy consumption and memory overhead. However, in Internet of Things (IoT) scenarios, memory and energy of nodes are scarce resources. Therefore, in NDN-based IoT applications, traditional caching strategies can cause node failures due to energy depletion, which can significantly reduce the network operational lifetime as well as create problems that caching is supposed to solve. In this paper, a caching strategy based on node centrality and energy availability, called Energy-aware Approximate Betweenness Centrality (EABC) is proposed for NDN-based IoT. EABC uses a topology-based heuristic to cache data content on nodes with high centrality and makes caching decisions based on the remaining energy of the nodes. We evaluate EABC using simulations based on ndnSIM in different topologies and compare it with several existing NDN caching strategies. The results show that EABC performs better in different types of network topologies, reduces the average transmission delay of data and balances the energy consumption of highly central nodes, thus extending the network lifetime.

关键词： Wireless communication Energy consumption Network topology Memory management Topology Delays Internet of Things

Genetic Auto-prompt Learning for Pre-trained Code Intelligence Language Models

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Feng, Chengzhe Sun, Yanan Li, Ke Zhou, Pan Lv, Jiancheng Lu, Aojun The College of Computer Science Sichuan University Chengdu610065 China The Department of Computer Science University of Exeter ExeterEX4 4QF United Kingdom The Hubei Engineering Research Center on Big Data Security School of Cyber Science and Engineering Huazhong University of Science and Technology Wuhan430074 China

As Pre-trained Language Models (PLMs), a popular approach for code intelligence, continue to grow in size, the computational cost of their usage has become prohibitively expensive. Prompt learning, a recent development in the field of natural language processing, emerges as a potential solution to address this challenge. In this paper, we investigate the effectiveness of prompt learning in code intelligence tasks. We unveil its reliance on manually designed prompts, which often require significant human effort and expertise. Moreover, we discover existing automatic prompt design methods are very limited to code intelligence tasks due to factors including gradient dependence, high computational demands, and limited applicability. To effectively address both issues, we propose Genetic Auto Prompt (GenAP), which utilizes an elaborate genetic algorithm to automatically design prompts. With GenAP, non-experts can effortlessly generate superior prompts compared to meticulously manual-designed ones. GenAP operates without the need for gradients or additional computational costs, rendering it gradient-free and cost-effective. Moreover, GenAP supports both understanding and generation types of code intelligence tasks, exhibiting great applicability. We conduct GenAP on three popular code intelligence PLMs with three canonical code intelligence tasks including defect prediction, code summarization, and code translation. The results suggest that GenAP can effectively automate the process of designing prompts. Specifically, GenAP outperforms all other methods across all three tasks (e.g. improving accuracy by an average of 2.13% for defect prediction). To the best of our knowledge, GenAP is the first work to automatically design prompts for code intelligence PLMs. Copyright © 2024, The Authors. All rights reserved.

关键词： Design

SpreadFGL: Edge-Client Collaborative Federated Graph Learning with Adaptive Neighbor Generation

学校读者我要写书评

暂无评论

SpreadFGL: Edge-Client Collaborative Federated Graph Learnin...

IEEE Annual Joint Conference: INFOCOM, IEEE computer and Communications Societies

作者： Luying Zhong Yueyang Pi Zheyi Chen Zhengxin Yu Wang Miao Xing Chen Geyong Min College of Computer and Data Science Fuzhou University China Fujian Provincial Key Laboratory of Network Computing and Intelligent Information Processing Fuzhou University China Engineering Research Center of Big Data Intelligence Ministry of Education China School of Computing and Communications University of Lancaster UK School of Engineering Computing and Mathematics University of Plymouth UK Department of Computer Science University of Exeter UK

ISBN: (数字)9798350383508

ISBN: (纸本)9798350383515

Federated Graph Learning (FGL) has garnered widespread attention by enabling collaborative training on multiple clients for semi-supervised classification tasks. However, most existing FGL studies do not well consider the missing inter-client topology information in real-world scenarios, causing insufficient feature aggregation of multi-hop neighbor clients during model training. Moreover, the classic FGL commonly adopts the FedAvg but neglects the high training costs when the number of clients expands, resulting in the overload of a single edge server. To address these important challenges, we propose a novel FGL framework, named SpreadFGL, to promote the information flow in edge-client collaboration and extract more generalized potential relationships between clients. In SpreadFGL, an adaptive graph imputation generator incorporated with a versatile assessor is first designed to exploit the potential links between subgraphs, without sharing raw data. Next, a new negative sampling mechanism is developed to make SpreadFGL concentrate on more refined information in downstream tasks. To facilitate load balancing at the edge layer, SpreadFGL follows a distributed training manner that enables fast model convergence. Using real-world testbed and benchmark graph datasets, extensive experiments demonstrate the effectiveness of the proposed SpreadFGL. The results show that SpreadFGL achieves higher accuracy and faster convergence against state-of-the-art algorithms.

关键词： Training Accuracy Collaboration Benchmark testing Imputation Generators Topology

An Auto-Parallel Method for Deep Learning Models Based on Genetic Algorithm 29

学校读者我要写书评

暂无评论

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

29th IEEE International Conference on Parallel and Distributed Systems, ICPADS 2023

作者： Zeng, Yan Huang, Chengchuang Ni, Yijie Zhou, Chunbao Zhang, Jilin Wang, Jue Zhou, Mingyao Xue, Meiting Zhang, Yunquan Hangzhou Dianzi University School of Computer Science and Technology Hangzhou310018 China Ministry of Education Key Laboratory for Modeling and Simulation of Complex Systems Hangzhou310018 China Data Security Governance Zhejiang Engineering Research Center Hangzhou310018 China Hangzhou Dianzi University School of ITMO Joint Institute Hangzhou310018 China Institute of Computer Network Information Center of the Chinese Academy of Sciences Beijing100086 China HuaWei China Institute of Computing Technology of the Chinese Academy of Sciences State Key Laboratory of Computer Architecture Beijing100086 China

ISBN: (纸本)9798350330717

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. © 2023 IEEE.

关键词： Auto-Parallel Distributed machine learning Genetic algorithm Integer linear programming Model parallelism

Global Context Aggregation Network for Lightweight Saliency Detection of Surface Defects

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Yan, Feng Jiang, Xiaoheng Lu, Yang Cui, Lisha Li, Shupan Cao, Jiale Xu, Mingliang Tao, Dacheng School of Computer Science and Artificial Intelligence Zhengzhou University Zhengzhou China Engineering Research Center of Intelligent Swarm Systems Ministry of Education Zhengzhou China National Supercomputing Center in Zhengzhou Zhengzhou China School of Electrical and Information Engineering Tianjin University Tianjin China the Sydney AI Centre the School of Computer Science Faculty of Engineering The University of Sydney DarlingtonNSW2008 Australia

Surface defect inspection is a very challenging task in which surface defects usually show weak appearances or exist under complex backgrounds. Most high-accuracy defect detection methods require expensive computation and storage overhead, making them less practical in some resource-constrained defect detection applications. Although some lightweight methods have achieved real-time inference speed with fewer parameters, they show poor detection accuracy in complex defect scenarios. To this end, we develop a Global Context Aggregation Network (GCANet) for lightweight saliency detection of surface defects on the encoder-decoder structure. First, we introduce a novel transformer encoder on the top layer of the lightweight backbone, which captures global context information through a novel Depthwise Self-Attention (DSA) module. The proposed DSA performs element-wise similarity in channel dimension while maintaining linear complexity. In addition, we introduce a novel Channel Reference Attention (CRA) module before each decoder block to strengthen the representation of multi-level features in the bottom-up path. The proposed CRA exploits the channel correlation between features at different layers to adaptively enhance feature representation. The experimental results on three public defect datasets demonstrate that the proposed network achieves a better trade-off between accuracy and running efficiency compared with other 17 state-of-the-art methods. Specifically, GCANet achieves competitive accuracy (91.79% Fβw, 93.55% Sα, and 97.35% E) on SD-saliency-900 while running 272fps on a single gpu. Copyright © 2023, The Authors. All rights reserved.

关键词： Defects