检索结果-内蒙古大学图书馆

Excitement surfeited turns to errors: Deep learning testing framework based on excitable neurons

INFORMATION SCIENCES 2023年第1期637卷

作者： Jin, Haibo Chen, Ruoxi Zheng, Haibin Chen, Jinyin Cheng, Yao Yu, Yue Chen, Tieming Liu, Xianglong Zhejiang Univ Technol Coll Informat Engn Hangzhou Peoples R China Zhejiang Univ Technol Inst Cyberspace Secur Hangzhou Peoples R China Huawei Int Singapore Singapore Natl Univ Def Technol Coll Comp Natl Lab Parallel & Distributed Proc Changsha Peoples R China Zhejiang Univ Technol Coll Comp Sci & Technol Hangzhou Peoples R China Beihang Univ State Key Lab Software Dev Environm Beijing Peoples R China

Despite impressive capabilities and outstanding performance, deep neural networks (DNNs) have become a growing concern among the public due to their frequently-appearing erroneous behaviors. As a result, it is urgent to test the security issues with DNNs systematically before they are deployed to the real world. Fine-grained metrics based on neuron coverage have been provided by existing testing approaches, and various approaches have been proposed to enhance these metrics. However, it has been gradually realized that a higher neuron coverage does not necessarily represent better capabilities in identifying defects that lead to errors. Besides, coverage-guided methods cannot hunt errors due to faulty training procedure. So the robustness improvement of DNNs via retraining by these testing examples are unsatisfactory. To tackle this challenge, we introduce the concept of excitable neurons based on Shapley value and design a white-box testing framework for DNNs, namely DeepSensor. It is motivated by our observation that neurons with larger responsibility towards model loss changes due to small perturbations are more likely related to incorrect corner cases due to potential defects. By maximizing the number of excitable neurons that correspond to various incorrect behaviors of models, DeepSensor can generate testing examples that effectively trigger more erroneous security issues caused by malicious inputs (both adversarial and polluted data) and incomplete training. Extensive experiments implemented on both image classification models and speaker recognition models have demonstrated the superiority of DeepSensor. Compared to the state-of-the-art testing methods, DeepSensor has the ability to find more wrong model behaviors due to malicious inputs (i.e., similar to x1.2 for adversarial and similar to x4.7 for polluted data) and incompletely-trained DNNs. Additionally, it can help DNNs build larger l(2)-norm robustness bound (similar to x3) via retraining according to C

关键词： Deep neural networks Deep learning testing Cooperative game theory Excitable neurons

来源：评论

学校读者我要写书评

暂无评论

Feasibility analysis of the Sanchez-Pomraning method to treat the particle size distribution in particle-dispersed fuel

引用

NUCLEAR ENGINEERING AND DESIGN 2024年 428卷

作者： Qin, Shuai Zhang, Qian Feng, Xing Liang, Yuechao Zhejiang Univ Zhejiang Inst Modern Phys Sch Phys Lab Adv Nucl Energy Theory & Applicat Hangzhou 310027 Zhejiang Peoples R China Huaneng Shandong Shidao Bay Nucl Power Co Ltd Rongcheng 264300 Shandong Peoples R China Natl Univ Def Technol Lab Digitizing Software Frontier Equipment Changsha 410073 Hunan Peoples R China Natl Univ Def Technol Natl Key Lab Parallel & Distributed Comp Changsha 410073 Hunan Peoples R China

The non-uniform size of particles appears in the particle-dispersed fuel generally, but this phenomenon is usually neglected. This study aims to evaluate the feasibility of the Sanchez-Pomraning method to solve the particle size distribution problem in particle-dispersed fuel. The Pu-rich agglomerate phenomenon in the MOX fuel and manufacturing tolerance in the Fully Ceramic Microencapsulated (FCM) fuel are calculated by ALPHA-DH code, which implements the Sanchez-Pomraning method based on the MOC transport calculation and improved subgroup resonance calculation. Numerical results show that the Sanchez-Pomraning method has the capability to treat the particle size distribution effect accurately, where the differences between the Sanchez-Pomraning method and the Monte Carlo method are generally lower than 160 pcm. For the Pu-rich agglomerate phenomenon, the results show that ignoring the Pu-rich agglomerate effect introduces the reactivity difference of around 600 pcm, and ignoring the size distribution of the Pu agglomerate will cause the reactivity of up to 200 pcm in the extra-high or extra-low Pu enrichment. The more discrete size distribution and the smaller Pu-spots, the more significant influence will occur in the MOX fuel. For the FCM fuel, the influence brought by the TRISO particle size distribution can be ignored when only TRISO particles are filled. However, ignoring the size distribution when TRISO particles and poison particles are both added into the FCM fuel introduces the differences reaching 600-1000 pcm. These results demonstrate the importance of considering the particle size distribution for analyzing the MOX fuel and FCM fuel in practice. And the Sanchez-Pomraning method is reliable to analyze the effect of particle size distribution.

关键词： Double heterogeneous Particle size distribution Pu spot Fully ceramic microencapsulated fuel Sanchez-Pomraning method

来源：评论

学校读者我要写书评

暂无评论

When Game Becomes Life: The Creators and Spectators of Online Game Replays and Live Streaming

引用

ACM TRANSACTIONS ON MULTIMEDIA COMPUTING COMMUNICATIONS AND APPLICATIONS 2016年第4期12卷 47-47页

作者： Jia, Adele Lu Shen, Siqi Epema, Dick H. J. Iosup, Alexandru China Agr Univ Coll Informat & Elect Engn Beijing Peoples R China Natl Univ Def Technol Parallel & Distributed Proc Lab Changsha Hunan Peoples R China Natl Univ Def Technol Sch Comp Changsha Hunan Peoples R China Delft Univ Technol Software & Comp Technol Dept NL-2600 AA Delft Netherlands Natl Univ Def Technol Coll Comp Changsha Hunan Peoples R China

Online gaming franchises such as World of Tanks, Defense of the Ancients, and StarCraft have attracted hundreds of millions of users who, apart from playing the game, also socialize with each other through gaming and viewing gamecasts. As a form of User Generated Content (UGC), gamecasts play an important role in user entertainment and gamer education. They deserve the attention of both industrial partners and the academic communities, corresponding to the large amount of revenue involved and the interesting research problems associated with UGC sites and social networks. Although previous work has put much effort into analyzing general UGC sites such as YouTube, relatively little is known about the gamecast sharing sites. In this work, we provide the first comprehensive study of gamecast sharing sites, including commercial streaming-based sites such as Amazon's *** and community-maintained replay-based sites such as WoTreplays. We collect and share a novel dataset on WoTreplays that includes more than 380,000 game replays, shared by more than 60,000 creators with more than 1.9 million gamers. Together with an earlier published dataset on ***, we investigate basic characteristics of gamecast sharing sites, and we analyze the activities of their creators and spectators. Among our results, we find that (i) WoTreplays and *** are both fast-consumed repositories, with millions of gamecasts being uploaded, viewed, and soon forgotten;(ii) both the gamecasts and the creators exhibit highly skewed popularity, with a significant heavy tail phenomenon;and (iii) the upload and download preferences of creators and spectators are different: while the creators emphasize their individual skills, the spectators appreciate team-wise tactics. Our findings provide important knowledge for infrastructure and service improvement, for example, in the design of proper resource allocation mechanisms that consider future gamecasting and in the tuning of incentive policies t

关键词： Online game communities gamecast sharing sites repository characteristics popularity dynamics user behaviors

来源：评论

学校读者我要写书评

暂无评论

A compressed matrix sequence method for solving normal equations of bundle adjustment

引用

MACHINE VISION AND APPLICATIONS 2021年第4期32卷 81-81页

作者： Peng, Jiaxin Liu, Jie Wei, Hua Natl Univ Def Technol Sch Comp Sci Lab Software Engn Complex Syst Changsha 410073 Hunan Peoples R China Natl Univ Def Technol Sch Comp Sci Parallel & Distributed Proc Lab Changsha 410073 Hunan Peoples R China Hubei Univ Sci & Technol Sch Resources Environm Sci & Engn Water & Land Resources Res Ctr Middle Reaches Yan Xianning 437100 Hubei Peoples R China Changsha Mouri Technol Co Ltd Algorithm Lab Changsha 410073 Hunan Peoples R China

Bundle adjustment is a least squares method-based algorithm for minimizing the global reprojection error and has provided an effective solution for structure from motion (SfM). The Levenberg-Marquardt algorithm provides a feasible and convenient way for bundle adjustment and creates a system of linear equations which are normal equations. For the special sparsity structure of an augmented Hessian matrix, a Schur complement trick is introduced to reduce computation complexity. However, the general sparse matrix storage formats are not optimized for the augmented Hessian matrix and consume too much computation time. According to the Schur complement trick, this paper divides the arrow-like augmented Hessian matrix into a structure matrix, a camera matrix and an observation matrix, and then proposes a new compressed matrix sequence (CMS) method to reduce time complexity for matrix operations. Under the definition of CMS, all of the matrices are stored in a dense form to accelerate the matrix operations, of which the matrix operations are redefined as well. CMS costs little computation time to build sparse matrices or access sparse matrices. The experimental results show that CMS achieves a significant speedup over general sparse matrix storage formats. Also, CMS being insensitive to the data input is more stable.

关键词： Structure from motion Bundle adjustment Levenberg-Marquardt algorithm

来源：评论

学校读者我要写书评

暂无评论

Document-level relation extraction with three channels

引用

KNOWLEDGE-BASED SYSTEMS 2024年 284卷

作者： Zhang, Zhanjun Zhao, Shan Zhang, Haoyu Wan, Qian Liu, Jie Natl Univ Def Technol Coll Comp Changsha Peoples R China Natl Univ Def Technol Sci & Technol Parallel & Distributed Proc Lab Changsha Peoples R China Natl Univ Def Technol Lab Software Engn Complex Syst Changsha Peoples R China Def Innovat Inst Beijing Peoples R China Hefei Univ Technol Sch Comp & Informat Hefei Peoples R China Natl Univ Def Technol 109 Deya Rd Changsha Peoples R China

Relation extraction aims to identify the relation between two entities. Early works on relation extraction are performed within a sentence. Numerous entity pairs with relations are distributed across sentences, so the task of document-level relation extraction is proposed. The current works perform well on document-level relation extraction through graph neural network, transformer, etc. However, these works extract relation features from one perspective, resulting in limited extraction performance of the model. To break through the limitation, this paper proposes the DRETC model consisting of three channels. Channel one first constructs the mention matrix and then extracts the relation features between entities. Channel two performs sequence inference from mention to entity before generating the relation matrix. Moreover, the bridge method is proposed to introduce information in channel one that is beneficial to relation extraction. Channel three directly generates entity representations through text sequences to make up for the absence of text information related to the entity. The three channels extract relation features from three different perspectives, and the combination leads to better extraction performance. Experimental results show that the proposed DRETC model outperforms previous strong baseline models with one channel on the public dataset DocRED. Further analysis demonstrates that the combination of multiple channels has a better relation extraction performance than a single channel.

关键词： Document-level relation extraction Three channels Relation features Relation matrix

来源：评论

学校读者我要写书评

暂无评论

PINN-FFHT: A physics-informed neural network for solving fluid flow and heat transfer problems without simulation data

引用

INTERNATIONAL JOURNAL OF MODERN PHYSICS C 2022年第12期33卷 2250166-2250166页

作者： Zhang, Qingyang Guo, Xiaowei Chen, Xinhai Xu, Chuanfu Liu, Jie Natl Univ Def Technol Sci & Technol Parallel & Distributed Proc Lab Changsha 410073 Hunan Peoples R China Natl Univ Def Technol Inst Quantum Informat Changsha 410073 Hunan Peoples R China Natl Univ Def Technol State Key Lab High Performance Comp Changsha 410073 Hunan Peoples R China Natl Univ Def Technol Lab Software Engn Complex Syst Changsha 410073 Hunan Peoples R China

In recent years, physics-informed neural networks (PINNs) have come to the foreground in many disciplines as a new way to solve partial differential equations. Compared with traditional discrete methods and data-driven surrogate models, PINNs can learn the solutions of partial differential equations without relying on tedious mesh generation and simulation data. In this paper, an original neural network structure PINN-FFHT based on PINNs is devised to solve the fluid flow and heat transfer problems. PINN-FFHT can simultaneously predict the flow field and take into consideration the influence of flow on the temperature field to solve the energy equation. A flexible and friendly boundary condition (BC) enforcement method and a dynamic strategy that can adaptively balance the loss term of velocity and that of temperature are proposed for training PINN-FFHT, serving to accelerate the convergence and improve the accuracy of predicted results. Three cases are predicted to validate the capabilities of the network, including the laminar flow with the Dirichlet BCs in heat transfer, respectively, under the Cartesian and the cylindrical coordinate systems, and the thermally fully developed flow with the Neumann BCs in heat transfer. Results show that PINN-FFHT is faster in convergence speed and higher in accuracy than traditional PINN methods.

关键词： Physics-informed neural networks (PINNs) partial differential equations fluid flow heat transfer boundary conditions

来源：评论

学校读者我要写书评

暂无评论

Feature and Performance Comparison of FaaS Platforms

Feature and Performance Comparison of FaaS Platforms

引用

IEEE International Conference on software Engineering and Service Sciences (ICSESS)

作者： Penghui Ma Peichang Shi Guodong Yi National Key Laboratory of Parallel and Distributed Processing College of Computer Science National University of Defense Technology Changsha China Key Laboratory of Software Engineering for Complex Systems College of Computer Science National University of Defense Technology Changsha China Xiangjiang Lab Changsha China School of Advanced Interdisciplinary Studies Hunan University Of Technology and Business Changsha China

With serverless computing offering more efficient and cost-effective application deployment, the diversity of serverless platforms presents challenges to users, including platform lock-in and costly migration. Moreover, due to the black box nature of function computing, traditional performance benchmarking methods are not applicable, necessitating new studies. This article presents a detailed comparison of six major public cloud function computing platforms and introduces a benchmarking framework for function computing performance. This framework aims to help users make comprehensive comparisons and select the most suitable platform for their specific needs.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Feedback Chain Network For Hippocampus Segmentation

arXiv

引用

arXiv 2022年

作者： Huang, Heyu Cong, Runmin Yang, Lianhe Du, Ling Wang, Cong Kwong, Sam School of Computer Science and Technology Tiangong University Tianjin300387 China The Institute of Information Science Beijing Jiaotong University Beijing100044 China The Distributed and Parallel Software Lab Huawei Technologies Shenzhen518129 China The Department of Computer Science City University of Hong Kong Hong Kong The City University of Hong Kong Shenzhen Research Institute Shenzhen51800 China

The hippocampus plays a vital role in the diagnosis and treatment of many neurological disorders. Recent years, deep learning technology has made great progress in the field of medical image segmentation, and the performance of related tasks has been constantly refreshed. In this paper, we focus on the hippocampus segmentation task and propose a novel hierarchical feedback chain network. The feedback chain structure unit learns deeper and wider feature representation of each encoder layer through the hierarchical feature aggregation feedback chains, and achieves feature selection and feedback through the feature handover attention module. Then, we embed a global pyramid attention unit between the feature encoder and the decoder to further modify the encoder features, including the pair-wise pyramid attention module for achieving adjacent attention interaction and the global context modeling module for capturing the long-range knowledge. The proposed approach achieves state-of-the-art performance on three publicly available datasets, compared with existing hippocampus segmentation approaches. The code and results can be found from the link of https://***/easymoneysniper183/sematic_seg. © 2022, CC BY-NC-SA.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

Three-Dimensional MT Conductive Anisotropic and Magnetic Modeling Using A - φ Potentials Employing a Mixed Nodal and Edge-Based Element Method

引用

APPLIED SCIENCES-BASEL 2024年第19期14卷 9019页

作者： Zhou, Zongyi Yi, Mingkuan Zhou, Junjun Cheng, Lianzheng Song, Tao Gong, Chunye Yang, Bo Xiao, Tiaojie Natl Univ Def Technol Lab Digitizing Software Frontier Equipment Changsha 410073 Peoples R China Natl Def Technol Sci & Technol Parallel & Distributed Proc Lab Changsha 410073 Peoples R China Henan Polytech Univ Dept Phys & Elect Informat Jiaozuo 454000 Peoples R China Kunming Univ Sch Math Kunming 650500 Peoples R China Guiyang Univ Coll Math & Informat Sci Guiyang 550005 Peoples R China Natl Supercomp Ctr Tianjin Tianjin 300450 Peoples R China

Magnetotelluric (MT) sounding is a geophysical technique widely utilized in mineral resource surveys, where conductivity and magnetic permeability serve as essential physical parameters for forward modeling and inversion. However, the effects of conductive anisotropy and non-zero magnetic susceptibility are usually ignored. In this study, we present a three-dimensional (3D) MT modeling algorithm using Coulomb-gauged electromagnetic potentials, incorporating a mixed nodal and edge-based finite element method capable of simulating MT responses for conductive anisotropic and magnetic anomalies. Subsequently, the algorithm's accuracy was validated in two steps: first, it was compared with analytical solutions for a 1D magnetic model;then, a comparison was made with previously published numerical results for a 3D generalized conductive anisotropic model. The results of two tests show that the maximum relative error is below 0.5% for both models. Furthermore, representative models were computed to comprehensively analyze the responses of MT. The findings illustrate the relationship between anisotropic parameters and electric fields and emphasize the significance of considering the impact of magnetic susceptibility in magnetite-rich regions.

关键词： magnetotelluric anisotropy three-dimensional magnetic forward modeling finite element method Coulomb-gauged potentials

来源：评论

学校读者我要写书评

暂无评论

FCloudless: A Performance-Aware Collaborative Mechanism for JointCloud Serverless

FCloudless: A Performance-Aware Collaborative Mechanism for ...

引用

IEEE International Conference on Joint Cloud Computing (JCC)

作者： Jianfei Liu Huaimin Wang Peichang Shi Yaojie Li Penghui Ma Guodong Yi National Key Laboratory of Parallel and Distributed Computing College of Computer Science National University of Defense Technology Changsha 410073 China Key Laboratory of Software Engineering for Complex Systems College of Computer Science National University of Defense Technology Changsha 410073 China Xiangjiang Lab Changsha 410073 China School of Advanced Interdisciplinary Studies Hunan University Of Technology and Business Changsha 410073 China

As a new stage in the development of the cloud computing paradigm, serverless computing has the high-level abstraction characteristic of shielding underlying details. This makes it extremely challenging for users to choose a suitable serverless platform. To address this, targeting the jointcloud computing scenario of heterogeneous serverless platforms across multiple clouds, this paper presents a jointcloud collaborative mechanism called FCloudless with cross-cloud detection of the full lifecycle performance of serverless platforms. Based on the benchmark metrics set that probe performance critical stages of the full lifecycle, this paper proposes a performance optimization algorithm based on detected performance data that takes into account all key stages that affect the performance during the lifecycle of a function and predicts the overall performance by combining the scores of local stages and dynamic weights. We evaluate FCloudless on AWS, AliYun, and Azure. The experimental results show that FCloudless can detect the underlying performance of serverless platforms hidden in the black box and its optimization algorithm can select the optimal scheduling strategy for various applications in a jointcloud environment. FCloudless reduces the runtime by 23.3% and 24.7% for cold and warm invocations respectively under cost constraints.

关键词：

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：