检索结果-内蒙古大学图书馆

23rd USENIX Conference on File and Storage Technologies, FAST 2025

作者： Chen, Weijian He, Shuibing Qu, Haoyang Zhang, Xuechen The State Key Laboratory of Blockchain and Data Security Zhejiang University China Zhejiang Lab China Institute of Blockchain and Data Security China Zhejiang Key Laboratory of Big Data Intelligent Computing China Washington State University Vancouver United States

ISBN: (纸本)9781939133458

Distributed training of graph neural networks (GNNs) has become a crucial technique for processing large graphs. Prevalent GNN frameworks are model-centric, necessitating the transfer of massive graph vertex features to GNN models, which leads to a significant communication bottleneck. Recognizing that the model size is often significantly smaller than the feature size, we propose LeapGNN, a feature-centric framework that reverses this paradigm by bringing GNN models to vertex features. To make it truly effective, we first propose a micrograph-based training strategy that leverages a refined structure to enhance locality, combined with the model migration technique, to minimize remote feature retrieval. Then, we devise a feature pre-gathering approach that merges multiple fetch operations into a single one to eliminate redundant feature transmissions. Finally, we employ a micrograph-based merging method that adjusts the number of micrographs for each worker to minimize kernel switches and synchronization overhead. Our experimental results demonstrate that LeapGNN achieves a performance speedup of up to 4.2× compared to the state-of-the-art method, namely P3 © 2025 FAST. All Rights Reserved.

关键词： Graph neural networks

来源：评论

学校读者我要写书评

暂无评论

GoPIM: GCN-Oriented Pipeline Optimization for PIM Accelerators 31

GoPIM: GCN-Oriented Pipeline Optimization for PIM Accelerato...

引用

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

作者： Yang, Siling He, Shuibing Wang, Wenjiong Yin, Yanlong Wu, Tong Chen, Weijian Zhang, Xuechen Sun, Xian-He Feng, Dan The State Key Laboratory of Blockchain and Data Security Zhejiang University China Zhejiang Lab China Institute of Blockchain and Data Security China Zhejiang Key Laboratory of Big Data Intelligent Computing China Washington State University Vancouver United States Illinois Institute of Technology United States Huazhong University of Science and Technology China Wuhan National Laboratory for Optoelectronics China

ISBN: (纸本)9798331506476

Graph convolutional networks (GCNs) are popular for a variety of graph learning tasks. ReRAM-based processing-in-memory (PIM) accelerators are promising to expedite GCN training owing to their in-situ computing capability. However, existing accelerators can be severely underutilized even with pipelines, due to the oversight of the skewed execution times of various GCN stages and the ignorance of skewed degrees of graph vertices. In this work, we propose GOPIM, a GCN-oriented pipeline optimization for PIM accelerators to expedite GCN training. First, GOPIM proposes an ML-based scheme that allocates crossbar resources to the most needed stages to streamline the overall pipeline. Second, GOPIM utilizes a selective vertex updating technique that evenly distributes vertices on crossbars by interleaved mapping. These techniques collectively reduce the overall execution time without losing much accuracy. We also provide a practical architecture design for GOPIM. Our experimental results show that, GoPIM achieves up to 191 × speedup and 16.1 × energy saving, compared to the state-of-the-art work. © 2025 IEEE.

关键词： Graph neural networks

来源：评论

学校读者我要写书评

暂无评论

OmniThink: Expanding Knowledge Boundaries in Machine Writing through Thinking

arXiv

引用

arXiv 2025年

作者： Xi, Zekun Yin, Wenbiao Fang, Jizhan Wu, Jialong Fang, Runnan Zhang, Ningyu Jiang, Yong Xie, Pengjun Huang, Fei Chen, Huajun Zhejiang University China Tongyi Lab Alibaba Group China Zhejiang Key Laboratory of Big Data Intelligent Computing China

Machine writing with large language models often relies on retrieval-augmented generation. However, these approaches remain confined within the boundaries of the model’s predefined scope, limiting the generation of content with rich information. Specifically, vanilla-retrieved information tends to lack depth, novelty, and suffers from redundancy, which negatively impacts the quality of generated articles, leading to shallow, unoriginal, and repetitive outputs. To address these issues, we propose OmniThink, a slow-thinking machine writing framework that emulates the human-like process of iterative expansion and reflection. The core idea behind OmniThink is to simulate the cognitive behavior of learners as they slowly deepen their knowledge of the topics. Experimental results demonstrate that OmniThink improves the knowledge density of generated articles without compromising metrics such as coherence and depth. Human evaluations and expert feedback further highlight the potential of OmniThink to address real-world challenges in the generation of long-form articles. Copyright © 2025, The Authors. All rights reserved.

关键词： Adversarial machine learning

来源：评论

学校读者我要写书评

暂无评论

LeapGNN: accelerating distributed GNN training leveraging feature-centric model migration 25

LeapGNN: accelerating distributed GNN training leveraging fe...

引用

Proceedings of the 23rd USENIX Conference on File and Storage Technologies

作者： Weijian Chen Shuibing He Haoyang Qu Xuechen Zhang The State Key Laboratory of Blockchain and Data Security Zhejiang University and Zhejiang Lab and Hangzhou High-Tech Zone (Binjiang) Institute of Blockchain and Data Security and Zhejiang Key Laboratory of Big Data Intelligent Computing Washington State University Vancouver

ISBN: (纸本)9781939133458

关键词：

来源：评论

学校读者我要写书评

暂无评论

How Do LLMs Acquire New Knowledge? A Knowledge Circuits Perspective on Continual Pre-Training

arXiv

引用

arXiv 2025年

作者： Ou, Yixin Yao, Yunzhi Zhang, Ningyu Jin, Hui Sun, Jiacheng Deng, Shumin Li, Zhenguo Chen, Huajun Zhejiang University China Huawei Noah’s Ark Lab Canada National University of Singapore NUS-NCS Joint Lab Singapore Zhejiang Key Laboratory of Big Data Intelligent Computing China

Despite exceptional capabilities in knowledge-intensive tasks, Large Language Models (LLMs) face a critical gap in understanding how they internalize new knowledge, particularly how to structurally embed acquired knowledge in their neural computations. We address this issue through the lens of knowledge circuit evolution, identifying computational subgraphs that facilitate knowledge storage and processing. Our systematic analysis of circuit evolution throughout continual pre-training reveals several key findings: (1) the acquisition of new knowledge is influenced by its relevance to preexisting knowledge;(2) the evolution of knowledge circuits exhibits a distinct phase shift from formation to optimization;(3) the evolution of knowledge circuits follows a deep-to-shallow pattern. These insights not only advance our theoretical understanding of the mechanisms of new knowledge acquisition in LLMs, but also provide potential implications for improving continual pre-training strategies to enhance model performance Copyright © 2025, The Authors. All rights reserved.

关键词： Knowledge acquisition

来源：评论

学校读者我要写书评

暂无评论

GoPIM: GCN-Oriented Pipeline Optimization for PIM Accelerators

GoPIM: GCN-Oriented Pipeline Optimization for PIM Accelerato...

引用

IEEE Symposium on High-Performance Computer Architecture

作者： Siling Yang Shuibing He Wenjiong Wang Yanlong Yin Tong Wu Weijian Chen Xuechen Zhang Xian-He Sun Dan Feng The State Key Laboratory of Blockchain and Data Security Zhejiang University Zhejiang Lab Hangzhou High-Tech Zone (Binjiang) Institute of Blockchain and Data Security Zhejiang Key Laboratory of Big Data Intelligent Computing Washington State University Vancouver Illinois Institute of Technology Huazhong University of Science and Technology Wuhan National Laboratory for Optoelectronics

ISBN: (数字)9798331506476

ISBN: (纸本)9798331506483

关键词： Training Accuracy Graph convolutional networks Pipelines Energy conservation Computer architecture Machine learning Parallel processing Optimization

来源：评论

学校读者我要写书评

暂无评论

K-ON: Stacking Knowledge On the Head Layer of Large Language Model

arXiv

引用

arXiv 2025年

作者： Guo, Lingbing Zhang, Yichi Bo, Zhongpu Chen, Zhuo Sun, Mengshu Zhang, Zhiqiang Zhang, Wen Chen, Huajun College of Computer Science and Technology Zhejiang University China ZJU-Ant Group Joint Lab of Knowledge Graph China Ant Group China School of Software Technology Zhejiang University China Zhejiang Key Laboratory of Big Data Intelligent Computing China

Recent advancements in large language models (LLMs) have significantly improved various natural language processing (NLP) tasks. Typically, LLMs are trained to predict the next token, aligning well with many NLP tasks. However, in knowledge graph (KG) scenarios, entities are the fundamental units and identifying an entity requires at least several tokens. This leads to a granularity mismatch between KGs and natural languages. To address this issue, we propose K-ON, which integrates KG knowledge into the LLM by employing multiple head layers for next k-step prediction. K-ON can not only generate entity-level results in one step, but also enables contrastive loss against entities, which is the most powerful tool in KG representation learning. Experimental results show that K-ON outperforms state-of-the-art methods that incorporate text and even the other modalities. Copyright © 2025, The Authors. All rights reserved.

关键词： Knowledge graph

来源：评论

学校读者我要写书评

暂无评论

Incomplete data management： a survey

引用

Frontiers of Computer Science 2018年第1期12卷 4-25页

作者： Xiaoye MIAO Yunjun GAO Su GUO Wanqi LIU College of Computer Science Zhejiang University Hangzhou 310027 China The Key Lab of Big Data Intelligent Computing of Zhejiang Province Zhejiang University Hangzhou 310027 China

Incomplete data accompanies our life processes and covers almost all fields of scientific studies, as a result of delivery failure, no power of battery, accidental loss, etc. However, how to model, index, and query incomplete data in- curs big challenges. For example, the queries struggling with incomplete data usually have dissatisfying query results due to the improper incompleteness handling methods. In this pa- per, we systematically review the management of incomplete data, including modelling, indexing, querying, and handling methods in terms of incomplete data. We also overview sev- eral application scenarios of incomplete data, and summa- rize the existing systems related to incomplete data. It is our hope that this survey could provide insights to the database community on how incomplete data is managed, and inspire database researchers to develop more advanced processing techniques and tools to cope with the issues resulting from incomplete data in the real world.

关键词： incomplete data query processing indexing,application system

来源：评论

学校读者我要写书评

暂无评论

SMILE: A Cost-Effective System for Serving Massive Pretrained Language Models in the Cloud 23

SMILE: A Cost-Effective System for Serving Massive Pretraine...

引用

2023 ACM/SIGMOD International Conference on Management of data, SIGMOD 2023

作者： Wang, Jue Chen, Ke Shou, Lidan Jiang, Dawei Chen, Gang Key Lab of Intelligent Computing Based Big Data of Zhejiang Province Zhejiang University Hangzhou China

ISBN: (纸本)9781450395076

Deep learning models, particularly pre-trained language models (PLMs), have become increasingly important for a variety of applications that require text/language processing. However, these models are resource-intensive and often require costly hardware such as dedicated GPU servers. In response to this issue, we present SMILE, a novel prototype system for efficient deployment and management of such models in the cloud. Our goal is to build a cloud platform from which tenants can easily derive their own custom models, and rent PLM processors to run inference services on these models at reduced costs. To facilitate this, we present a co-design of cost-effective storage and computation scheme for managing massive customized PLMs with constrained hardware resources via effective resource sharing and multiplexing. Our system consists of four core components: vPLM creator, vPLM storage appliance, vPLM trainer, and vPLM processor, which allow tenants to easily create, store, train, and use their customized PLM in the cloud without the need for dedicated hardware or maintenance. In particular, vPLM processors are virtualized from a physical machine, and are designed to have a multi-tenant nature, enabling efficient utilization of resources by precomputing the intermediate representation of PLMs and using adapters to provide customization instead of training the entire model. This allows tenants to host their PLMs in the cloud at minor costs. In our demonstration, we show that over 10,000 models can be hosted on one single machine without compromising the inference speed and accuracy. Overall, our system provides a convenient and cost-effective solution for tenants to host and manage PLMs in the cloud for their customized tasks. © 2023 ACM.

关键词： Cost effectiveness

来源：评论

学校读者我要写书评

暂无评论

Dual Enhancement for Multi-label Learning with Missing labels 21

Dual Enhancement for Multi-Label Learning with Missing Label...

引用

4th International Conference on Machine Learning and Machine Intelligence, MLMI 2021

作者： Liu, Shengyuan Wang, Haobo Hu, Tianlei Chen, Ke Key Lab of Intelligent Computing Based Big Data of Zhejiang Province College of Computer Science and Technology Zhejiang University China

ISBN: (纸本)9781450384247

The goal of multi-label learning with missing labels (MLML) is assigning each testing instance multiple labels given training instances that have a partial set of labels. The most challenging issue is to complete the missing labels by leveraging input signals and given labels. Existing methods usually complete missing entries by utilizing label correlation, which, however, overlook knowledge from the object itself. To tackle this problem, we propose a dual enhancement framework that enhances missing entries from the label- perspective and instance-perspective. First, we develop a Long Short-Term Memory (LSTM) network that utilizes sequential label correlations to complete missing labels. Second, we expand the training dataset via a MixUp procedure to further enhance the label confidence. Extensive experiments on three multi-label image datasets clearly verify the superiority of our proposed method to the baselines. © 2021 ACM.

关键词： Image recognition

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：