检索结果-内蒙古大学图书馆

2022 IEEE International Conference on Multimedia and Expo, ICME 2022

作者： Ma, Bole Tong, Chao School of Computer Science and Engineering Beihang University China Beihang University State Key Laboratory of Software Development Environment China Yunnan Key Laboratory of Blockchain Application Technology China

ISBN: (数字)9781665485630

ISBN: (纸本)9781665485630

Multimodal networks that juxtapose visual and linguistic modalities are currently widely adopted for solving vision-and-language tasks. They perform well in simple and intu-itive tasks, but are prone to mistakes in tasks involving latent or implicit details, due to the difficulty of capturing crucial but imperceptible visual signals in the real world. Perception errors lead to nonsensical results, but can be corrected by commonsense knowledge. To this end, we combine visual perception and linguistic commonsense to solve the challenging daily events causality reasoning task. We propose a novel Object-Aware Reasoning Network to focus on object inter-action while ignoring distracting information to refine visual perception. Further, a language branch with an independent prediction head is supervised to learn causality commonsense to help correct obvious perception errors, resulting in more plausible conclusions. Extensive experiments demonstrate that our method achieves new state-of-the-art results on Vis-Causal dataset. © 2022 IEEE.

关键词： Linguistics

来源：评论

学校读者我要写书评

暂无评论

On Efficient Packet Batching and Resource Allocation for GPU based NFV Acceleration 31

On Efficient Packet Batching and Resource Allocation for GPU...

引用

31st IEEE/ACM International Symposium on Quality of Service, IWQoS 2023

作者： Zeng, Deze Zhu, Andong Gu, Lin Chen, Quan Guo, Minyi School of Computer Science China University of Geosciences Hubei Wuhan China Nanjing University State Key Lab. for Novel Software Technology Nanjing China School of Computer Science and Technology Huazhong University of Science and Technology Wuhan China Shanghai Jiao Tong University Department of Computer Science and Engineering Shanghai China

ISBN: (纸本)9798350399738

Network Function Virtualization (NFV) has already become an essential technology for improving the scalab.lity and flexibility of modern computer networks. The performance gap has become the main issue that impedes the development of NFV. GPUs, with massive parallel processors, are advocated to accelerate the Virtualized Network Functions (VNFs). However, the special architecture and workflow of GPUs introduce new challenges, especially on the batched processing, and resource allocation. In this paper, we propose GPU-based NFV Acceleration framework (GNFA) with an efficient packet batching and resource allocation solution. Considering the increased latency caused by the accumulation of the GPU kernel invoking overhead, we first invent a latency reduction mechanism called SM Performance Compensation (SPC). A Partition and Adjustment based Batching and Resource Allocation (PABARA) algorithm that jointly considers batch size tuning and GPU thread allocation is also proposed. We have practically implemented GNFA and extensively evaluated its performance on some well-known VNFs. The experiment results show that GNFA can effectively promote the GPU resource utilization and improve the NFV performance in terms of per-packet latency. © 2023 IEEE.

关键词： Graphics processing unit

来源：评论

学校读者我要写书评

暂无评论

Effective Training of PINNs by Combining CMA-ES with Gradient Descent

Effective Training of PINNs by Combining CMA-ES with Gradien...

引用

Congress on Evolutionary Computation

作者： Lin Liu Yuan Yuan School of Computer Science and Engineering Beihang University Beijing China State Key Laboratory of Software Development Environment Zhongguancun Laboratory School of Computer Science and Engineering Beihang University Beijing China

ISBN: (数字)9798350308365

ISBN: (纸本)9798350308372

Physics-Informed Neural Networks (PINNs) have recently received increasing attention, however, optimizing the loss function of PINNs is notoriously difficult, where the landscape of the loss function is often highly non-convex and rugged. Local optimization methods based on gradient information can converge quickly but are prone to being trapped in local minima for training PINNs. Evolutionary algorithms (EAs) are well known for the global search ability, which can help escape from local minima. It has been reported in the literature that EAs show some advantages over gradient-based methods in training PINNs. Inspired by the Memetic Algorithm, we combine global-search based EAs and local-search based batch gradient descent in order to make the best of both word. In addition, since the PINN loss function is composed of multiple terms, balancing these terms is also a challenging issue. Therefore, we also attempt to combine EAs with multiple-gradient descent algorithm for multi-objective optimization. Our experiments provide strong evidence for the superiority of the above algorithms.

关键词： Training Memetics Neural networks Evolutionary computation Pareto optimization Prediction algorithms Linear programming

来源：评论

学校读者我要写书评

暂无评论

Understanding heterophily for graph neural networks 24

Understanding heterophily for graph neural networks

引用

Proceedings of the 41st International Conference on Machine Learning

作者： Junfu Wang Yuanfang Guo Liang Yang Yunhong Wang State Key Laboratory of Software Development Environment and School of Computer Science and Engineering and Shen Yuan Honors College Beihang University Beijing China State Key Laboratory of Software Development Environment and School of Computer Science and Engineering Beihang University Beijing China School of Artificial Intelligence Hebei University of Technology Tianjin China School of Computer Science and Engineering Beihang University Beijing China

Graphs with heterophily have been regarded as challenging scenarios for Graph Neural Networks (GNNs), where nodes are connected with dissimilar neighbors through various patterns. In this paper, we present theoretical understandings of heterophily for GNNs by incorporating the graph convolution (GC) operations into fully connected networks via the proposed Heterophilous Stochastic Block Models (HSBM), a general random graph model that can accommodate diverse heterophily patterns. Our theoretical investigation comprehensively analyze the impact of heterophily from three critical aspects. Firstly, for the impact of different heterophily patterns, we show that the separability gains are determined by two factors, i.e., the Euclidean distance of the neighborhood distributions and $\sqrt{\mathbb{E}\left[\operatorname{deg}\right]}$, where $\mathbb{E}\left[\operatorname{deg}\right]$ is the averaged node degree. Secondly, we show that the neighborhood inconsistency has a detrimental impact on separability, which is similar to degrading $\mathbb{E}\left[\operatorname{deg}\right]$ by a specific factor. Finally, for the impact of stacking multiple layers, we show that the separability gains are determined by the normalized distance of the l-powered neighborhood distributions, indicating that nodes still possess separability in various regimes, even when over-smoothing occurs. Extensive experiments on both synthetic and realworld data verify the effectiveness of our theory.

关键词：

来源：评论

学校读者我要写书评

暂无评论

A Framework to Analyze the Impact of Maps on VANET Simulation

引用

Chinese Journal of Electronics 2023年第3期20卷 401-406页

作者： Xiaoming LI Jingxuan WANG Xin JIN Tong ZHAO Wei YAN State Key Laboratory of Software Development Environment Beihang University Beijing China School of Electronic Engineering and Computer Science Peking University Beijing China

Mobility models play an important role on VANET simulation, which has been discussed in many studies. However, few studies have analyzed the impact of maps. We find that maps can greatly influence simulation results. We present a framework to help researchers analyze the impact of maps. We regard intersections in the map as virtual nodes and propose Virtual node coverage (VNC) and Virtual connected components (VCC) as two metrics to estimate the impact of maps. Through the-oretical analysis and extensive simulations, we show that our approach can be applied to various scenarios and researchers can use our approach to analyze the impact of maps on VANET simulation.

关键词： Measurement Analytical models Electric potential Mobility models Simulation Vehicular ad hoc networks Research and development

来源：评论

学校读者我要写书评

暂无评论

Pooling Under the Sun: A Mining Pool Centralized Revisit and Solution

Pooling Under the Sun: A Mining Pool Centralized Revisit and...

引用

IEEE International Conference on Trust, Security and Privacy in Computing and Communications (TrustCom)

作者： Kundu Chen Jie Luo State Key Laboratory of Software Development Environment School of Computer Science and Engineering Beihang University Beijing China

ISBN: (数字)9798350381993

ISBN: (纸本)9798350382006

Mining pools are fundamental infrastructure in most public blockchain networks that utilize Proof-of-Work consensus, ensuring that miners receive stable rewards and contribute to the computational power and storage of the network. However, mining pools also introduce centralization to the network, which increases the risk of security issues such as bribe attacks and censorship. To address the issue, this paper proposes a decentralized mining pool schema SSMP. Firstly, we distinguish between two aspects of centralization in mining pools, which are data determination, and block ownership, and suggest solutions to mitigate these issues. The proposed method utilizes anonymous broadcasting for a random leader selection schema to ensure each participating miner has an equal possibility to generate the block data. Additionally, a special threshold signature design prevents the leader miner from controlling the valid block solely of the mining pool through bribery. To validate our approach, we theoretically verify the security of the proposed method and compare it to traditional centralized mining pools. In summary, SSMP withholds the essential properties of the decentralized mining pools that maintain efficiency.

关键词： Privacy Voting Proof of Work Censorship Blockchains Safety Encryption

来源：评论

学校读者我要写书评

暂无评论

TRANSFORMER-PATCHER: ONE MISTAKE WORTH ONE NEURON 11

TRANSFORMER-PATCHER: ONE MISTAKE WORTH ONE NEURON

引用

11th International Conference on Learning Representations, ICLR 2023

作者： Huang, Zeyu Shen, Yikang Zhang, Xiaofeng Zhou, Jie Rong, Wenge Xiong, Zhang State Key Laboratory of Software Development Environment Beihang University China Sino-French Engineer School Beihang University China School of Computer Science and Engineering Beihang University China Mila University of Montreal Canada WeChat AI Tencent Inc. China

Large Transformer-based Pretrained Language Models (PLMs) dominate almost all Natural Language Processing (NLP) tasks. Nevertheless, they still make mistakes from time to time. For a model deployed in an industrial environment, fixing these mistakes quickly and robustly is vital to improve user experiences. Previous works formalize such problems as Model Editing (ME) and mostly focus on fixing one mistake. However, the one-mistake-fixing scenario is not an accurate abstraction of the real-world challenge. In the deployment of AI services, there are ever-emerging mistakes, and the same mistake may recur if not corrected in time. Thus a preferable solution is to rectify the mistakes as soon as they appear nonstop. Therefore, we extend the existing ME into Sequential Model Editing (SME) to help develop more practical editing methods. Our study shows that most current ME methods could yield unsatisfying results in this scenario. We then introduce Transformer-Patcher, a novel model editor that can shift the behavior of transformer-based models by simply adding and training a few neurons in the last Feed-Forward Network layer. Experimental results on both classification and generation tasks show that Transformer-Patcher can successively correct up to thousands of errors (Reliability) and generalize to their equivalent inputs (Generality) while retaining the model's accuracy on irrelevant inputs (Locality). Our method outperforms previous fine-tuning and HyperNetwork-based methods and achieves state-of-the-art performance for Sequential Model Editing (SME). The code is availab.e at https://***/ZeroYuHuang/Transform er-Patcher. © 2023 11th International Conference on Learning Representations, ICLR 2023. All rights reserved.

关键词： Network layers

来源：评论

学校读者我要写书评

暂无评论

Towards Fully Decoupled End-to-End Person Search

Towards Fully Decoupled End-to-End Person Search

引用

2023 International Conference on Digital Image Computing: Techniques and Applications, DICTA 2023

作者： Zhang, Pengcheng Bai, Xiao Zheng, Jin Ning, Xin Jiangxi Research Institute Beihang University School of Computer Science and Engineering State Key Laboratory of Software Development Environment Beijing China Institute of Semiconductors Chinese Academy of Sciences Cognitive Computing Technology Joint Laboratory Wave Group Beijing China

ISBN: (纸本)9798350382204

End-to-end person search aims to jointly detect and re-identify a target person in raw scene images with a unified model. The detection task unifies all persons while the re-id task discriminates different identities, resulting in conflict optimal objectives. Existing works proposed to decouple end-to-end person search to alleviate such conflict. Yet these methods are still sub-optimal on one or two of the sub-tasks due to their partially decoupled models, which limits the overall person search performance. In this paper, we propose to fully decouple person search towards optimal person search. A task-incremental person search network is proposed to incrementally construct an end-to-end model for the detection and re-id sub-task, which decouples the model architecture for the two sub-tasks. The proposed task-incremental network allows task-incremental training for the two conflicting tasks. This enables independent learning for different objectives thus fully decoupled the model for personsearch. Comprehensive experimental evaluations demonstrate the effectiveness of the proposed fully decoupled models for end-to-end person search. © 2023 IEEE.

关键词： computer vision

来源：评论

学校读者我要写书评

暂无评论

Approximate k-Nearest Neighbor Query over Spatial Data Federation 28th

Approximate k-Nearest Neighbor Query over Spatial Data Fede...

引用

28th International Conference on Database Systems for Advanced Applications, DASFAA 2023

作者： Zhang, Kaining Tong, Yongxin Shi, Yexuan Zeng, Yuxiang Xu, Yi Chen, Lei Zhou, Zimu Xu, Ke Lv, Weifeng Zheng, Zhiming State Key Laboratory of Software Development Environment Beijing Advanced Innovation Center for Future Blockchain and Privacy Computing School of Computer Science and Engineering and Institute of Artificial Intelligence Beihang University Beijing China Department of Computer Science and Engineering HKUST Hong Kong Guangzhou China City University of Hong Kong Hong Kong

ISBN: (纸本)9783031306365

Approximate nearest neighbor query is a fundamental spatial query widely applied in many real-world applications. In the big data era, there is an increasing demand to scale these queries over a spatial data federation, which consists of multiple data owners, each holding a private, disjoint partition of the entire spatial dataset. However, it is non-trivial to enable approximate k-nearest neighbor query over a spatial data federation. This is because stringent security constraints are often imposed to protect the sensitive, privately owned data partitions, whereas naively extending prior secure query processing solutions leads to high inefficiency (e.g., 100 s per query). In this paper, we propose two novel algorithms for efficient and secure approximate k-nearest neighbor query over a spatial data federation. We theoretically analyze their communication cost and time complexity, and further prove their security guarantees and approximation bounds. Extensive experiments show that our algorithms outperform the state-of-the-art solutions with respect to the query efficiency and often yield a higher accuracy. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Nearest neighbor search

来源：评论

学校读者我要写书评

暂无评论

CODESWIFT: Accelerating LLM Inference for Efficient Code Generation

arXiv

引用

arXiv 2025年

作者： Zhao, Qianhui Zhang, Li Liu, Fang Lian, Xiaoli Meng, Qiaoyuanhe Jiao, Ziqian Zhou, Zetong Zhang, Borui Guo, Runlin Li, Jia School of Computer Science and Engineering State Key Laboratory of Complex & Critical Software Environment Beihang University China Key Lab of High Confidence Software Technology Peking University China

Code generation is a latency-sensitive task that demands high timeliness, but the autoregressive decoding mechanism of Large Language Models (LLMs) leads to poor inference efficiency. Existing LLM inference acceleration methods mainly focus on standalone functions using only built-in components. Moreover, they treat code like natural language sequences, ignoring its unique syntax and semantic characteristics. As a result, the effectiveness of these approaches in code generation tasks remains limited and fails to align with real-world programming scenarios. To alleviate this issue, we propose CODESWIFT, a simple yet highly efficient inference acceleration approach specifically designed for code generation, without comprising the quality of the output. CODESWIFT constructs a multi-source datastore, providing access to both general and project-specific knowledge, facilitating the retrieval of high-quality draft sequences. Moreover, CODESWIFT reduces retrieval cost by controlling retrieval timing, and enhances efficiency through parallel retrieval and a context- and LLM preference-aware cache. Experimental results show that CODESWIFT can reach up to 2.53× and 2.54× speedup compared to autoregressive decoding in repository-level and standalone code generation tasks, respectively, outperforming state-of-the-art inference acceleration approaches by up to 88%. Our code and data are availab.e at https://***/r/CodeSwift. Copyright © 2025, The Authors. All rights reserved.

关键词： Decoding

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：