检索结果-内蒙古大学图书馆

AccelES: Accelerating Top-K SpMV for Embedding Similarity via Low-bit Pruning

学校读者我要写书评

暂无评论

AccelES: Accelerating Top-K SpMV for Embedding Similarity vi...

IEEE Symposium on High-Performance Computer Architecture

作者： Jiaqi Zhai Xuanhua Shi Kaiyi Huang Chencheng Ye Weifang Hu Bingsheng He Hai Jin National Engineering Research Center for Big Data Technology and System/Services Computing Technology and System Lab/Cluster and Grid Computing Lab School of Computer Science and Technology Huazhong University of Science and Technology Wuhan China School of Computing National University of Singapore Singapore

ISBN: (数字)9798331506476

ISBN: (纸本)9798331506483

In the realm of recommendation systems, achieving real-time performance in embedding similarity tasks is often hindered by the limitations of traditional Top-K sparse matrix-vector multiplication (SpMV) methods, which suffer from high latency due to inefficient memory access patterns. This paper identifies these critical gaps and introduces AccelES, a novel approach that significantly enhances the efficiency of Top-K SpMV. Our method employs a two-stage calculation scheme: the first stage utilizes a compact, low-bit dataset to quickly identify the most relevant entries, while the second stage performs full-precision calculations solely on this pruned subset, thereby minimizing computational overhead. Furthermore, AccelES incorporates innovative matrix representations, Ultra-CSR and Random-CSR, which optimize memory bandwidth utilization. Experimental results demonstrate that AccelES accelerates performance, surpassing state-of-the-art FPGA, GPU, and CPU solutions by factors of 3.4×, 2.5×, and 153.3×, respectively, under controlled conditions. These advancements not only enhance processing speed but also significantly improve real-time performance in recommendation systems, establishing AccelES as a pivotal contribution to the field of Top-K sparse matrix-vector multiplication.

关键词： Quantization (signal) High dimensional data Memory management Graphics processing units Bandwidth Real-time systems Sparse matrices Recommender systems Field programmable gate arrays Optimization

OUTLIER SYNTHESIS VIA HAMILTONIAN MONTE CARLO FOR OUT-OF-DISTRIBUTION DETECTION

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Li, Hengzhuang Zhang, Teng National Engineering Research Center for Big Data Technology and System Service Computing Technology and Systems Laboratory Cluster and Grid Computing Lab School of Computer Science and Technology Huazhong University of Science and Technology Wuhan China

Out-of-distribution (OOD) detection is crucial for developing trustworthy and reliable machine learning systems. Recent advances in training with auxiliary OOD data demonstrate efficacy in enhancing detection capabilities. Nonetheless, these methods heavily rely on acquiring a large pool of high-quality natural outliers. Some prior methods try to alleviate this problem by synthesizing virtual outliers but suffer from either poor quality or high cost due to the monotonous sampling strategy and the heavy-parameterized generative models. In this paper, we overcome all these problems by proposing the Hamiltonian Monte Carlo Outlier Synthesis (HamOS) framework, which views the synthesis process as sampling from Markov chains. Based solely on the in-distribution data, the Markov chains can extensively traverse the feature space and generate diverse and representative outliers, hence exposing the model to miscellaneous potential OOD scenarios. The Hamiltonian Monte Carlo with sampling acceptance rate almost close to 1 also makes our framework enjoy great efficiency. By empirically competing with SOTA baselines on both standard and large-scale benchmarks, we verify the efficacy and efficiency of our proposed HamOS. Our code is available at: https://***/Fir-lat/HamOS_OOD. © 2025, CC BY.

关键词： Markov chains

How to Select Pre-Trained Code Models for Reuse? A Learning Perspective

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Bi, Zhangqian Wan, Yao Chu, Zhaoyang Hu, Yufei Zhang, Junyi Zhang, Hongyu Xu, Guandong Jin, Hai National Engineering Research Center for Big Data Technology and System Services Computing Technology and System Lab Cluster and Grid Computing Lab Wuhan China School of Computer Science and Technology Huazhong University of Science and Technology Wuhan China School of Big Data and Software Engineering Chongqing University Chongqing China School of Computer Science University of Technology Sydney Sydney Australia

Pre-training a language model and then fine-tuning it has shown to be an efficient and effective technique for a wide range of code intelligence tasks, such as code generation, code summarization, and vulnerability detection. However, pre-training language models on a large-scale code corpus is computationally expensive. Fortunately, many off-the-shelf Pre-trained Code Models (PCMs), such as CodeBERT, CodeT5, CodeGen, and Code Llama, have been released publicly. These models acquire general code understanding and generation capability during pre-training, which enhances their performance on downstream code intelligence tasks. With an increasing number of these public pre-trained models, selecting the most suitable one to reuse for a specific task is essential. In this paper, we systematically investigate the reusability of PCMs. We first explore three intuitive model selection methods that select by size, training data, or brute-force fine-tuning. Experimental results show that these straightforward techniques either perform poorly or suffer high costs. Motivated by these findings, we explore learning-based model selection strategies that utilize pre-trained models without altering their parameters. Specifically, we train proxy models to gauge the performance of pre-trained models, and measure the distribution deviation between a model's latent features and the task's labels, using their closeness as an indicator of model transferability. We conduct experiments on 100 widely-used open-source PCMs for code intelligence tasks, with sizes ranging from 42.5 million to 3 billion parameters. The results demonstrate that learning-based selection methods reduce selection time to 100 seconds, compared to 2,700 hours with brute-force fine-tuning, with less than 6% performance degradation across related tasks. © 2025, CC BY-NC-SA.

关键词： Adversarial machine learning

ACGraph: Accelerating Streaming Graph Processing via Dependence Hierarchy 23

学校读者我要写书评

暂无评论

ACGraph: Accelerating Streaming Graph Processing via Depende...

Proceedings of the 60th Annual ACM/IEEE Design Automation Conference

作者： Zihan Jiang Fubing Mao Yapu Guo Xu Liu Haikun Liu Xiaofei Liao Hai Jin Wei Zhang National Engineering Research Center for Big Data Technology and System Services Computing Technology and System Lab Cluster and Grid Computing Lab School of Computer Science and Technology Huazhong University of Science and Technology Wuhan China Department of Electronic and Computer Engineering HKUST Hong Kong

ISBN: (纸本)9798350323481

Streaming graph processing needs to timely evaluate continuous queries. Prior systems suffer from massive redundant computations due to the irregular order of processing vertices influenced by updates. To address this issue, we propose ACGraph, a novel streaming graph processing approach for monotonic graph algorithms. It maintains dependence trees during runtime, and makes affected vertices processed in a top-to-bottom order in the hierarchy of the dependence trees, thus normalizing the state propagation order and coalescing of multiple propagation to the same vertices. Experimental results show that ACGraph reduces the number of updates by 50% on average, and achieves the speedup of 1.75~7.43× over state-of-the-art systems.

关键词： graph processing

MeHyper: Accelerating Hypergraph Neural Networks by Exploring Implicit dataflows 31

学校读者我要写书评

暂无评论

MeHyper: Accelerating Hypergraph Neural Networks by Explorin...

31st IEEE International Symposium on High Performance Computer Architecture, HPCA 2025

作者： Zhao, Wenju Yao, Pengcheng Chen, Dan Zheng, Long Liao, Xiaofei Wang, Qinggang Ma, Shaobo Li, Yu Liu, Haifeng Xiao, Wenjing Sun, Yufei Zhu, Bing Jin, Hai Xue, Jingling Huazhong University of Science and Technology National Engineering Research Center for Big Data Technology and System Services Computing Technology and System Lab Cluster and Grid Computing Lab School of Computer Science and Technology Wuhan430074 China National University of Singapore School of Computing 119077 Singapore Guangxi University School of Computer Electronics and Information NanNing530004 China University of New South Wales School of Computer Science and Engineering SydneyNSW2052 Australia

ISBN: (纸本)9798331506476

Hypergraph Neural Networks (HGNNs) are increasingly utilized to analyze complex inter-entity relationships. Traditional HGNN systems, based on a hyperedge-centric dataflow model, independently process aggregation tasks for hyperedges and vertices, leading to significant computational redundancy. This redundancy arises from recalculating shared information across different tasks. For the first time, we identify and harness implicit dataflows (i.e., dependencies) within HGNNs, introducing the microedge concept to effectively capture and reuse intricate shared information among aggregation tasks, thereby minimizing redundant computations. We have developed a new microedge-centric dataflow model that processes shared information as fine-grained microedge aggregation tasks. This dataflow model is supported by the Read-Process-Activate-Generate execution model, which aims to optimize parallelism among these tasks. Furthermore, our newly developed MeHyper, a microedge-centric HGNN accelerator, incorporates a decoupled pipeline for improved computational parallelism and a hierarchical feature management strategy to reduce off-chip memory accesses for large volumes of intermediate feature vectors generated. Our evaluation demonstrates that MeHyper substantially outperforms the leading CPUbased system PyG-CPU and the GPU-based system HyperGef, delivering performance improvements of 1,032.23 × and 10.51 ×, and energy efficiencies of 1,169.03 × and 9.96 ×, respectively. © 2025 IEEE.

关键词： Redundancy

It Takes Two to Tango: Serverless Workflow Serving via Bilaterally Engaged Resource Adaptation

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Wu, Jing Wang, Lin Deng, Quanfeng Yu, Chen Zhang, Dong Yan, Bingheng Liu, Fangming National Engineering Research Center for Big Data Technology and System Services Computing Technology and System Lab Cluster and Grid Computing Lab Huazhong University of Science and Technology China Paderborn University Germany Inspur Data Co. Ltd. China Peng Cheng Laboratory China

Serverless platforms typically adopt an early-binding approach for function sizing, requiring developers to specify an immutable size for each function within a workflow beforehand. Accounting for potential runtime variability, developers must size functions for worst-case scenarios to ensure service-level objectives (SLOs), resulting in significant resource inefficiency. To address this issue, we propose Janus, a novel resource adaptation framework for serverless platforms. Janus employs a late-binding approach, allowing function sizes to be dynamically adapted based on runtime conditions. The main challenge lies in the information barrier between the developer and the provider: developers lack access to runtime information, while providers lack domain knowledge about the workflow. To bridge this gap, Janus allows developers to provide hints containing rules and options for resource adaptation. Providers then follow these hints to dynamically adjust resource allocation at runtime based on real-time function execution information, ensuring compliance with SLOs. We implement Janus and conduct extensive experiments with real-world serverless workflows. Our results demonstrate that Janus enhances resource efficiency by up to 34.7% compared to the state-of-the-art. © 2025, CC BY-NC-ND.

关键词： Resource allocation

Breakthrough in breast tumor detection and diagnosis: a noise-resilient, rotation-invariant framework

学校读者我要写书评

暂无评论

Multimedia Tools and Applications 2025年 1-27页

作者： Nosheen, Fariha Khan, Salabat Sharif, Muhammad Kim, Do Hyeun Alkanhel, Reem AbdelSamee, Nagwan Department of Computer Science COMSATS University Islamabad Attock Campus Attock43600 Pakistan Department of Computer Engineering Jeju National University Jeju Special Self-Governing Province Jeju-Si63243 Korea Republic of Big Data Research Center Jeju National University Jeju63243 Korea Republic of Department of Information Technology College of Computer and Information Sciences Princess Nourahbint Abdulrahman University P.O. Box 84428 Riyadh11671 Saudi Arabia

The fact that early-stage breast cancer typically presents no signs, poses a global risk to the lives of women. Digital mammography is just one method among many that can detect breast cancer in its early stages. Despite extensive research, most methods for detecting breast cancers still produce a large number of false positives. The difficulty in improving detection accuracy is in reducing false positives by differentiating masses from normal tissues. Using the textural properties of the masses, this study aims to develop a computer-aided diagnosis system that reduces the number of false positive and negative mammography results. The suggested method initially partitions regions of interest (ROI) into small patches to extract an abnormality region and micro-pattern, which permits the extraction of specific information about the image's content in targeted areas. Reducing the computational complexity of an image analysis operation is accomplished by partitioning a ROI into smaller patches rather than processing the complete image at once. Due to the significant impact the noise has on detection accuracy in mammograms, a textural descriptor that is insensitive to noise is introduced to describe image features such as lines, spots, flat areas, and edges. By expanding the rotation-invariant and noise-tolerant descriptor histograms, we generate an improved histogram that preserves the original regional patterns and spatial relationships between masses. To distinguish between "normal" and "mass" and "benign" and "malignant", Support Vector Machines (SVM) is employed with grid search based hyperparameter optimization. Our proposed approach was evaluated using the Digital database for Screening Mammography (DDSM), which contains over 1024 ROI cases. This database is a well-established benchmark for testing new mammography analysis methods. Each case in the DDSM database has been analyzed and interpreted by qualified radiologists, with comprehensive information provided thr

关键词： Diseases

FedEAT: A Robustness Optimization Framework for Federated LLMs

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Pang, Yahao Wu, Xingyuan Zhang, Xiaojin Chen, Wei Jin, Hai National Engineering Research Center for Big Data Technology and System Services Computing Technology and System Lab Cluster and Grid Computing Lab School of Computer Science and Technology Huazhong University of Science and Technology Wuhan430074 China Huazhong University of Science and Technology Wuhan China

Significant advancements have been made by Large Language Models (LLMs) in the domains of natural language understanding and automated content creation. However, they still face persistent problems, including substantial computational costs and inadequate availability of training data. The combination of Federated Learning (FL) and LLMs (federated LLMs) offers a solution by leveraging distributed data while protecting privacy, which positions it as an ideal choice for sensitive domains. However, Federated LLMs still suffer from robustness challenges, including data heterogeneity, malicious clients, and adversarial attacks, which greatly hinder their applications. We first introduce the robustness problems in federated LLMs, to address these challenges, we propose FedEAT (Federated Embedding space Adversarial Training), a novel framework that applies adversarial training in the embedding space of client LLM and employs a robust aggregation approach, specifically geometric median aggregation, to enhance the robustness of Federated LLMs. Our experiments demonstrate that FedEAT effectively improves the robustness of Federated LLMs with minimal performance loss. © 2025, CC BY.

关键词： Federated learning

SPViT: Accelerate Vision Transformer Inference on Mobile Devices via Adaptive Splitting and Offloading

学校读者我要写书评

暂无评论

IEEE Transactions on Mobile Computing 2025年

作者： Zhao, Sifan Liu, Tongtong Jin, Hai Yao, Dezhong Huazhong University of Science and Technology National Engineering Research Center for Big Data Technology and System Services Computing Technology and System Lab Cluster and Grid Computing Lab School of Computer Science and Technology Wuhan430074 China

The Vision Transformer (ViT), which benefits from utilizing self-attention mechanisms, has demonstrated superior accuracy compared to CNNs. However, due to the expensive computational costs, deploying and inferring ViTs on resource-constrained mobile devices has become a challenge. To resolve this challenge, we conducted an empirical analysis to identify performance bottlenecks in deploying ViTs on mobile devices and explored viable solutions. In this paper, we propose SPViT, an adaptive split and offloading method that accelerates ViT inference on mobile devices. SPViT executes collaborative inference of ViT across available edge devices. We introduce a fine-grained splitting technique for the vision transformer structure. Furthermore, we propose an algorithm based on the Auto Regression model to predict partition latency and adaptive offload partitions. Finally, we design offline and online optimization methods to minimize the computational and communication overhead on each device. Based on real-world prototype experiments, SPViT effectively reduces inference latency by 2.2x to 3.3x across four state-of-the-art models. © 2002-2012 IEEE.

关键词： Regression analysis