检索结果-内蒙古大学图书馆

27th European Conference on Artificial Intelligence, ECAI 2024

作者： Li, Wenda Chen, Kaixuan Liu, Shunyu Zheng, Tongya Huang, Wenjie Song, Mingli State Key Laboratory of Blockchain and Security Zhejiang University China School of Software Technology Zhejiang University China Institute of Blockchain and Data Security China Big Graph Center School of Computer and Computing Science Hangzhou City University China College of Computer Science and Technology Zhejiang University Hangzhou China

ISBN: (纸本)9781643685489

Mini-batch Graph Transformer (MGT), as an emerging graph learning model, has demonstrated significant advantages in semi-supervised node prediction tasks with improved computational efficiency and enhanced model robustness. However, existing methods for processing local information either rely on sampling or simple aggregation, which respectively result in the loss and squashing of critical neighbor information. Moreover, the limited number of nodes in each mini-batch restricts the model’s capacity to capture the global characteristic of the graph. In this paper, we propose LGMformer, a novel MGT model that employs a two-stage augmented interaction strategy, transitioning from local to global perspectives, to address the aforementioned bottlenecks. The local interaction augmentation (LIA) presents a neighbor-target interaction Transformer (NTIformer) to acquire an insightful understanding of the co-interaction patterns between neighbors and the target node, resulting in a locally effective token list that serves as input for the MGT. In contrast, global interaction augmentation (GIA) adopts a cross-attention mechanism to incorporate entire graph prototypes into the target node representation, thereby compensating for the global graph information to ensure a more comprehensive perception. To this end, LGMformer achieves the enhancement of node representations under the MGT paradigm. Experimental results related to node classification on the ten benchmark datasets demonstrate the effectiveness of the proposed method. Our code is available at https://***/l-wd/LGMformer. © 2024 The Authors.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

PancakeFS: A Write Efficiently and Read Optimized Filesystem

PancakeFS: A Write Efficiently and Read Optimized Filesystem

引用

Consumer Electronics and computer Engineering (ICCECE), 2021 IEEE International Conference on

作者： Yongkang Wang Yang Liu Zhiguang Chen School of Computer Science and Engineering Sun Yat-sen University Guangzhou China State Key Laboratory of High-Performance Computing Institute for Quantum Information College of Computer National University of Defense Technology Changsha China

With the development of emerging technologies such as cloud computing and large AI models (such as LLM), many applications have placed higher demands on the intensive read and write processing of massive data. Addressing the limitations of traditional file systems, we propose PancakeFS, an LSM-Tree-based file system. Leveraging the write-friendly characteristics of LSM-Tree, PancakeFS exhibits efficient write performance. We improve read performance through the design of an inode allocation algorithm based on subtree partitioning and optimization of the disk layout for hot data reconstruction. Our experimental results demonstrate that PancakeFS outperforms Ext4 and TableFS in terms of both read and write performance.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Precise and scalable evaluation on the robustness metric of neural networks as classifiers 2

Precise and scalable evaluation on the robustness metric of ...

引用

2nd International Conference on Electronics, Communications and Information technology, CECIT 2021

作者： Hu, Yangyi Institute for Quantum Information State Key Laboratory of High Performance Computing National University of Defense Technology College of Computer Science and Technology Changsha410073 China

ISBN: (纸本)9781665437578

It is well known that the minimum adversarial distortion associated with a specific sample x0 reflects the local robustness of neural networks. However, it is intractable to solve the optimization problem related to the minimum adversarial distortion for general neural network cause it is non-convex. Works have studied the lower bound or upper bound of the minimum adversarial distortion to give a robustness metrics for neural networks, such as CLEVER score and CW attack. In this paper, we provide a formal robustness guarantee to transform the robustness analysis into two sub-problems: (1), compute out the maximum of the objective function gt* (rt);(2), generate a sequence {rth} s.t. limn→∞ rth is the minimum adversarial distortion for targeted attack. Based on this transformation, we propose an efficient and effective algorithm to directly estimate the instance-specific minimum adversarial distortion on the norm of the input manipulation required to change the classifier decision. Experimental results on two data-sets, MNIST and CIFAR, show that our measure of robustness is between the lower bound given by CLEVER and the upper bound by CW, indicating our approach gives a precise measure of the robustness of neural network. In addition, the powerful mathematical technique Extreme Value Estimation enables our algorithm computationally feasible for large neural networks. To the best of our knowledge, our algorithm is the first attack-independent approach to directly evaluate the minimum adversarial distortion as a robustness metric for neural network. © 2021 IEEE.

关键词： Estimation

来源：评论

学校读者我要写书评

暂无评论

ReSLB: Load Balanced Workflow for Distributed Deep Learning Mass Spectrometry Database 24

ReSLB: Load Balanced Workflow for Distributed Deep Learning ...

引用

24th IEEE International Conference on high performance computing and Communications, 8th IEEE International Conference on Data science and Systems, 20th IEEE International Conference on Smart City and 8th IEEE International Conference on Dependability in Sensor, Cloud and Big Data Systems and Application, HPCC/DSS/SmartCity/DependSys 2022

作者： Mocheng, Li Yang, Liu Yang, Ou Zhiguang, Chen Nong, Xiao Tao, Chen College of Computer National University of Defense Technology Institute for Quantum Information State Key Laboratory of High Performance Computing Changsha China College of Computer National University of Defense Technology Changsha China School of Computer Sun Yat-sen University Guangzhou China Beijing China

ISBN: (纸本)9798350319934

The proteomics data analysis pipeline based on the shotgun method requires efficient data processing methods. The parallel algorithm of mass spectrometry database search faces the problems of rapidly expanding database size but low scalability. The heterogeneous database search algorithm based on deep learning is an effective way to solve this problem. Still, the deep learning-based distributed parallel database search algorithm is lacking. This paper analyzes the database searching computational load using a neural network and designs ReSLB, a workflow for the restricted search of mass spectrometry data based on GPU cluster and neural network scoring algorithm. This work aims to ensure the high scalability of future deep learning-based mass spectrometry distributed databases. In the performance estimation of 256 GPUs, the load imbalance of less than 30% and the parallel efficiency of 60% are achieved. Compared with state-of-the-art, the time cost is reduced by 75%, and the parallel efficiency from 1 to 256 GPUs is 3.6x higher than that. © 2022 IEEE.

关键词： Mass spectrometry

来源：评论

学校读者我要写书评

暂无评论

A Low BER Cooperative-adaptive-equalizer for Serial Receiver in HPC Networks

A Low BER Cooperative-adaptive-equalizer for Serial Receiver...

引用

IEEE International Conference on high performance computing and Communications (HPCC)

作者： Lai Mingche Lv Fangxu Zhang Geng Xu Chaolong State Key Laboratory of High Performance Computing School of Computer National University of Defense Technology Changsha China

With the evolution of high-performance computer and data center interconnect communication rates from 53.125Gbps to 106.25Gbps, high-speed serial interfaces face a dramatic increase in BER challenges, which seriously affects interconnect performance and system stability. To address the challenge of high BER of 106.25Gbps PAM4 receiver, this paper proposes a cooperative adaptive equalizer structure, which adopts three equalizers and adaptive cooperative equalization algorithm to achieve low BER under large insertion loss conditions. In addition, a blind adaptive equalization algorithm based on judgment feedback equalizer is proposed to shorten the link training time. In this paper, a 12nm CMOS process is used to design a receiver based on cooperative adaptive equalizer. Simulation shows that the receiver with cooperative adaptive equalizer can support BER less than e-12 for a de-emphasized l06.25Gbps PAM4 signal after 34.3dB @ 26.5625G Hz channel, and the convergence period is less then 0.2 us.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Explicitly represent the boundary of ReLU NN as a classifier and measure its robustness 2

Explicitly represent the boundary of ReLU NN as a classifier...

引用

2nd International Conference on Electronics, Communications and Information technology, CECIT 2021

ISBN: (纸本)9781665437578

Adversarial examples pose a great threat to the application of neural network as a classifier in areas with high security requirements. Intuitively, the adversarial property of neural network is closely related to the boundary. However, works barely give a rigorous description of it. This paper for the first time studies the boundary of ReLU NN, which separates the sample space into distinct cells, each of which belongs to the same class. During the demonstration, we discuss the algebraic property of border points upon the separating boundary and the topological difference between border points and odd ones. We derive Theorem III.7 to give a rigorous, comprehensive and concise description on the boundary of ReLU NN. Furthermore, an algorithm, based on the description, is proposed to give a standard and exact metric for the local robustness of ReLU NN. This metric gives the local adversarial property of NN and reflects some aspects of its security performance, i.e. the nimimum distortion needed to craft an adversarial example. To the best of our knowledge, this is the first work proposed to give a rigorous and example-specific metric for the local robustness of ReLU NN. © 2021 IEEE.

关键词： computer vision

来源：评论

学校读者我要写书评

暂无评论

Cryogenic Control and Readout Integrated Circuits for Solid-state Quantum computing

arXiv

引用

arXiv 2024年

作者： Lei, Lingxiao Huang, Heng Chen, Pingxing Deng, Mingtang Institute for Quantum Information State Key Laboratory of High Performance Computing College of Computer Science and Technology National University of Defense Technology Changsha410073 China Institute for Quantum Science and Technology College of Sciences National University of Defense Technology Changsha410073 China

In the pursuit of quantum computing, solid-state quantum systems, particularly superconducting ones, have made remarkable advancements over the past two decades. However, achieving fault-tolerant quantum computing for next-generation applications necessitates the integration of several million qubits, which presents significant challenges in terms of interconnection complexity and latency that are currently unsolvable with state-of-the-art room-temperature control and readout electronics. Recently, cryogenic integrated circuits (ICs), including CMOS radio-frequency ICs and rapid-single-flux-quantum-logic ICs, have emerged as potential alternatives to room-temperature electronics. Unlike their room-temperature counterparts, these ICs are deployed within cryostats to enhance scalability by reducing the number and length of transmission lines. Additionally, operating at cryogenic temperatures can suppress electronic noise and improve qubit control fidelity. However, for CMOS ICs specifically, circuit design uncertainties arise due to a lack of reliable models for cryogenic field effect transistors as well as issues related to severe fickle noises and power dissipation at cryogenic temperatures. This paper provides a comprehensive review of recent research on both types of cryogenic control and readout ICs but primarily focuses on the more mature CMOS technology. The discussion encompasses principles underlying control and readout techniques employed in cryogenic CMOS ICs along with their architectural designs;characterization and modeling approaches for field effect transistors under cryogenic conditions;as well as fundamental concepts pertaining to rapid single flux quantum circuits. In conclusion, while prototypes of cryogenic ICs have been successfully fabricated recently, further studies are still necessary;we believe that rapid single-flux-quantum-based or hybrid ICs featuring enhanced digital components may represent the future direction in this emerging area. ©

关键词： CMOS integrated circuits

来源：评论

学校读者我要写书评

暂无评论

An Automata-Theoretic Approach to Synthesizing Binarized Neural Networks 21st

An Automata-Theoretic Approach to Synthesizing Binarized Ne...

引用

21st International Symposium on Automated technology for Verification and Analysis, ATVA 2023

作者： Tao, Ye Liu, Wanwei Song, Fu Liang, Zhen Wang, Ji Zhu, Hongxu College of Computer Science and Technology National University of Defense Technology Changsha China School of Information Science and Technology ShanghaiTech University Shanghai China Institute of Software Chinese Academy of Sciences and University of Chinese Academy of Sciences Beijing China Automotive Software Innovation Center Shanghai China Institute for Quantum Information and State Key Laboratory for High Performance Computing National University of Defense Technology Changsha China

ISBN: (纸本)9783031453281

Deep neural networks, (DNNs, a.k.a. NNs), have been widely used in various tasks and have been proven to be successful. However, the accompanied expensive computing and storage costs make the deployments in resource-constrained devices a significant concern. To solve this issue, quantization has emerged as an effective way to reduce the costs of DNNs with little accuracy degradation by quantizing floating-point numbers to low-width fixed-point representations. Quantized neural networks (QNNs) have been developed, with binarized neural networks (BNNs) restricted to binary values as a special case. Another concern about neural networks is their vulnerability and lack of interpretability. Despite the active research on trustworthy of DNNs, few approaches have been proposed to QNNs. To this end, this paper presents an automata-theoretic approach to synthesizing BNNs that meet designated properties. More specifically, we define a temporal logic, called BLTL, as the specification language. We show that each BLTL formula can be transformed into an automaton on finite words. To deal with the state-explosion problem, we provide a tableau-based approach in real implementation. For the synthesis procedure, we utilize SMT solvers to detect the existence of a model (i.e., a BNN) in the construction process. Notably, synthesis provides a way to determine the hyper-parameters of the network before training. Moreover, we experimentally evaluate our approach and demonstrate its effectiveness in improving the individual fairness and local robustness of BNNs while maintaining accuracy to a great extent. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

DSGA: Distractor-Suppressing Graph Attention for Multi-object Tracking 22

DSGA: Distractor-Suppressing Graph Attention for Multi-objec...

引用

Proceedings of the 8th International Conference on Robotics and Artificial Intelligence

作者： Guangjie Gao Yan Gao Liyang Xu Huibin Tan Yuhua Tang Institute for Quantum Information & State Key Laboratory of High Performance Computing College of Computer Science and Technology National University of Defense Technology China Chinese Institute of Electronics China

ISBN: (纸本)9781450397544

Multiple object tracking (MOT) methods based on single object tracking are of great interest because of their ability to balance efficiency and performance on the strength of the localization capability of single-target tracking. However, most of the single object tracking methods only distinguish foreground and background. They are susceptible to the influence of similar interfering objects during localization, while in multiple object tracking scenarios, there are more interfering objects and the influence is more severe. Therefore, we propose a Distractor-Suppressing Graph Attention (DSGA) to learn more discriminative attention by reducing the influence of distractors on learning attention weight features. Furthermore, DSGA is embedded into the basic MOT framework “SiamMOT” formed as DSGA-SiamMOT and applied to multiple object tracking to verify its effectiveness. We conduct experiments on the MOT Challenge benchmark with "public detection", and obtain MOTA 66.65%, IDF1 62.2% accuracy on the MOT17 dataset with 14fps.

关键词： graph attention mot sot distractor suppressing

来源：评论

学校读者我要写书评

暂无评论

Romat: Role-Based Multi-Agent Transformer for Generalizable Heterogeneous Cooperation

SSRN

引用

SSRN 2023年

作者： Wang, Dongzi Zhong, Fangwei Wen, Muning Li, Minglong Peng, Yuanxi Li, Teng Yang, Yaodong State Key Laboratory of High Performance Computing College of Computer Science and Technology National University of Defense Technology Changsha China Peking University Beijing China Shanghai China

Multi-task multi-agent systems (MASs) are challenging to model because they involve heterogeneous agents with different behavior patterns that need to cooperate across various tasks. Existing networks for single-agent policies are not suitable for this setting, as they cannot share policies among agents without losing task-specific performance. We propose a novel framework called Role-based Multi-Agent Transformer (RoMAT), which uses a sequence modeling technique and a role-based actor to enable agents to adapt to different tasks and roles in MASs. RoMAT has a modular model architecture, where backbone networks are shared by all agents, but a small part of the parameters (role-based actor) is independent, depending on the agents’ exclusive structures. We evaluate RoMAT on several benchmark tasks and show that it can capture the behavior patterns of heterogeneous agents and achieve better performance and generalization than other methods in both single and multi-task settings. © 2023, The Authors. All rights reserved.

关键词： Multi agent systems

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：