检索结果-内蒙古大学图书馆

1st Workshop on Gender-Inclusive Translation Technologies, GITT 2023

作者： Cabrera, Lena Niehues, Jan Department of Advanced Computing Sciences Maastricht University Netherlands Institute for Anthropomatics and Robotics Karlsruhe Institute of Technology Germany

ISBN: (纸本)9789464857177

Neural machine translation (NMT) models often suffer from gender biases that harm users and society at large. In this work, we explore how bridging the gap between languages for which parallel data is not available affects gender bias in multilingual NMT, specifically for zero-shot directions. We evaluate translation between grammatical gender languages which requires preserving the inherent gender information from the source in the target language. We study the effect of encouraging language-agnostic hidden representations on models’ ability to preserve gender and compare pivot-based and zero-shot translation regarding the influence of the bridge language (participating in all language pairs during training) on gender preservation. We find that language-agnostic representations mitigate zero-shot models’ masculine bias, and with increased levels of gender inflection in the bridge language, pivoting surpasses zero-shot translation regarding fairer gender preservation for speaker-related gender agreement. © 2023 The Authors.

关键词： Zero-shot learning

来源：评论

学校读者我要写书评

暂无评论

Task-Adaptive Generative Adversarial Network Based Speech Dereverberation for Robust Speech Recognition 1

引用

18th National Conference on Man-Machine Speech Communication, NCMMSC 2023

作者： Liu, Ji Li, Nan Ge, Meng Fu, Yanjie Wang, Longbiao Dang, Jianwu Tianjin Key Laboratory of Cognitive Computing and Application College of Intelligence and Computing Tianjin University Tianjin China Japan Advanced Institute of Science and Technology Ishikawa Nomi Japan

ISBN: (数字)9789819706013

ISBN: (纸本)9789819706006

Reverberation is known to severely affect speech recognition performance when speech is recorded in an enclosed space. Deep learning-based speech dereverberation has been remarkably successful in recent years, achieving superior recognition performance for far-field speech applications. However, the output from conventional dereverberation systems cannot be guaranteed suitable for back-end recognition systems because of their different task goals. To bridge the gap between the front-end dereverberation and the back-end recognition, we propose a novel task-adaptive speech dereverberation generative adversarial network (GAN) based speech dereverberation model called Task-adaptive GAN. Specifically, we propose to replace the binary-valued discriminator in a regular generative adversarial network with a novel senone-predicted discriminator, and also introduce a well-designed recognition-aware generator as a dereverberation system. By doing so, the corresponding output distribution will be more suitable for the recognition task. Experimental results on the REVERB corpus show that our proposed approach achieves a relative 18.6% and 8.6% word error rate reduction than the traditional GAN-based baseline system on the simulated set and real set, respectively. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Generative adversarial networks

来源：评论

学校读者我要写书评

暂无评论

Static Keystroke Dynamic Authentication (SKDA) Model to Authenticate User during Password Change 3

Static Keystroke Dynamic Authentication (SKDA) Model to Auth...

引用

3rd International Conference on Smart Data Intelligence, ICSMDI 2023

作者： Raul, Nataasha Shankarmani, Radha Joshi, Padmaja Sardar Patel Institute of Technology Mumbai India Centre for Development of Advanced Computing Mumbai India

ISBN: (纸本)9781665464871

Keystroke dynamics is considered as a supporting factor of authentication. Especially in the static keystroke dynamics, the user is identified by using the timing featured, which is captured while the user enters the login ID and password. To achieve this, the user profile needs to be created with timing features. However, in a scenario like a change of password where nearly no keystroke timing data is available, non-conventional features may be helpful. This article focuses on using non-conventional features such as NumLock key, Shift key, CapsLock key, etc for identifying users during a change of password. The paper also details how to capture the non-conventional features in static keystroke dynamics and build a model that can be used in the change of nassword. © 2023 IEEE.

关键词： Authentication

来源：评论

学校读者我要写书评

暂无评论

Gradient Kernelized Cat Boost Pre-Emptive Traffic Aware Data Distribution in Cloud computing

Gradient Kernelized Cat Boost Pre-Emptive Traffic Aware Data...

引用

2023 IEEE International Conference on Integrated Circuits and Communication Systems, ICICACS 2023

作者： Radha, S. Jeyalaksshmi, S. School of Computing Sciences Vels Institute of Science Technology and Advanced Studies Chennai India School of Computing Sciences Vels Institute of Science Chennai India

ISBN: (纸本)9798350398465

Cloud provides a distributed computing flexible on-demand data services for the individual person and organizations to store and access their data anywhere and anytime through the internet without user. Every day a large volume of data is generated from different sources. The large volume of data occupies excessive bandwidth and it becomes directed to network traffic. The traditional processing tools are unable to solve effectively for network traffic during the large volume of data transmission. Therefore, a novel technique called Gradient kernelized Cat Boost Priority Pre-emptive traffic aware data distribution (GECOPP) method is introduced for minimizing the traffic occurrence level in the cloud environment during the large volume of data distribution. First, the server receives the number of user-requested data from the user. Then the GECOPP method classifies the user-requested data into an emergency or normal based on different resource factors such as energy, bandwidth, arrival data size, and session time by using Nesterov Gradient kernelized Cat Boost classifier. The task assigner stores the classified data in the priority queue to minimize the traffic occurrence level with lesser space complexity. Finally, a pre-emptive approach is employed for distributing the data to multiple data centers at different locations hence it minimizing the traffic occurrence level in the cloud environment. Experimental evaluation of the GECOPP Method is carried out on the different factors such as classification accuracy, space complexity, makespan, and throughput with respect to the number of data and number of cloud users. The observed results indicate that the GECOPP technique offers an efficient solution in terms of achieving higher accuracy, and throughput, and minimizing the makespan as well as space complexity than the conventional techniques. © 2023 IEEE.

关键词： Cloud computing

来源：评论

学校读者我要写书评

暂无评论

Pressure-dependent electronic,optical,and mechanical properties of antiperovskite X_(3)NP(X=Ca,Mg):A first-principles study

引用

Journal of Semiconductors 2023年第10期44卷 52-60页

作者： Chunbao Feng Changhe Wu Xin Luo Tao Hu Fanchuan Chen Shichang Li Shengnan Duan Wenjie Hou Dengfeng Li Gang Tang Gang Zhang School of Science Chongqing University of Posts and TelecommunicationsChongqing 400065China Institute for Advanced Sciences Chongqing University of Posts and TelecommunicationsChongqing 400065China School of Computer Science and Technology Northwestern Polytechnical UniversityXi’an 710129China Advanced Research Institute of Multidisciplinary Science Beijing Institute of TechnologyBeijing 100081China Institute of High Performance Computing A*STAR138632Singapore

Hydrostatic pressure provides an efficient way to tune and optimize the properties of solid materials without chang-ing their *** this work,we investigate the electronic,optical,and mechanical properties of antiperovskite X_(3)NP(X^(2+)=Ca,Mg)upon compression by first-principles *** results reveal that the system is anisotropic,and the lat-tice constant a of X_(3)NP exhibits the fastest rate of decrease upon compression among the three directions,which is different from the typical Pnma phase of halide and chalcogenide ***,Ca_(3)NP has higher compressibility than Mg_(3)NP due to its small bulk *** electronic and optical properties of Mg_(3)NP show small fluctuations upon compression,but those of Ca_(3)NP are more sensitive to pressure due to its higher compressibility and lower unoccupied 3d orbital *** example,the band gap,lattice dielectric constant,and exciton binding energy of Ca_(3)NP decrease rapidly as the pressure *** addition,the increase in pressure significantly improves the optical absorption and theoretical conversion effi-ciency of Ca_(3)***,the mechanical properties of X_(3)NP are also increased upon compression due to the reduction in bond length,while inducing a brittle-to-ductile *** research provides theoretical guidance and insights for future experi-mental tuning of the physical properties of antiperovskite semiconductors by pressure.

关键词： antiperovskite hydrostatic pressure physical properties first-principles calculations

来源：评论

学校读者我要写书评

暂无评论

Strengthening-softening transition and maximum strength in Schwarz nanocrystals

引用

Nano Materials Science 2024年第3期6卷 320-328页

作者： Hanzheng Xing Jiaxi Jiang Yujia Wang Yongpan Zeng Xiaoyan Li Center for Advanced Mechanics and Materials Applied Mechanics LaboratoryDepartment of Engineering MechanicsTsinghua UniversityBeijing100084China Institute of High Performance Computing(IHPC) Agency for ScienceTechnology and Research(A*STAR)Singapore138632Singapore

Recently,a Schwarz crystal structure with curved grain boundaries(GBs)constrained by twin-boundary(TB)networks was discovered in nanocrystalline Cu through experiments and atomistic *** Cu with nanosized Schwarz crystals exhibited high strength and excellent thermal ***,the grainsize effect and associated deformation mechanisms of Schwarz nanocrystals remain ***,we performed large-scale atomistic simulations to investigate the deformation behaviors and grain-size effect of nanocrystalline Cu with Schwarz *** simulations showed that similar to regular nanocrystals,Schwarz nanocrystals exhibit a strengthening-softening transition with decreasing grain *** critical grain size in Schwarz nanocrystals is smaller than that in regular nanocrystals,leading to a maximum strength higher than that of regular *** simulations revealed that the softening in Schwarz nanocrystals mainly originates from TB migration(or detwinning)and annihilation of GBs,rather than GB-mediated processes(including GB migration,sliding and diffusion)dominating the softening in regular *** analyses of simulation data further showed that compared with those in regular nanocrystals,the GB-mediated processes in Schwarz nanocrystals are suppressed,which is related to the low volume fraction of amorphous-like GBs and constraints of TB *** smaller critical grain size arises from the suppression of GB-mediated processes.

关键词： Schwarz nanocrystal Curved grain boundary Atomistic simulation Grain size effect Maximum strength

来源：评论

学校读者我要写书评

暂无评论

What Keeps Women in IT Degrees 14

What Keeps Women in IT Degrees

引用

14th IEEE Integrated STEM Education Conference, ISEC 2024

作者： Molnar, Andreea Brüggemann-Klein, Anne Swinburne University of Technology Department of Computing Technologies Melbourne Australia Institute for Advanced Study Technical University of Munich Munich Germany

ISBN: (纸本)9798350352801

Despite increased efforts to attract and support women in IT fields, their numbers remain low. Several factors contribute to this, including the retention of women once they enroll in IT degrees. In this study, we shed light on how an initiative helps women in IT fields. Through interviews with women enrolled in these fields, we found that this initiative creates a space where they can find mentorship, develop a sense of belonging, and connect with role models. Additionally, personal characteristics like confidence, self-determination, and the ability to stand up for oneself were perceived by the women as useful in this field. © 2024 IEEE.

关键词： diversity gender higher education interventions IT retention role-models women

来源：评论

学校读者我要写书评

暂无评论

Framework for understanding intention-unbreakable malware

引用

Science China(Information Sciences) 2023年第4期66卷 174-192页

作者： Tiantian JI Binxing FANG Xiang CUI Zhongru WANG Peng LIAO Shouyou SONG Key Laboratory of Trustworthy Distributed Computing and Service(BUPT) Ministry of Education Beijing University of Posts and Telecommunications Cyberspace Institute of Advanced Technology Guangzhou University Chinese Academy of Cyberspace Studies Beijing Dig Apis Technology Co. Ltd.

The anti-analysis technology of malware has always been the focus in the cyberspace security field. As malware analysis techniques evolve, malware writers continually employ sophisticated anti-reverse engineering techniques to defeat and evade state-of-the-art analyzers. Therefore, to prepare for unknown attacks, studying malware analysis techniques is insufficient. More importantly, we should study new antianalysis techniques for malware. This paper expands the concept of anti-analysis malware and defines a type of intention-unbreakable malware(IUM). To help defenders fully understand and establish a defense system against the new security threats, this paper systematically discusses IUM from the perspectives of threat discovery, threat modeling, attribute analysis, and threat assessment; in addition, this study also provides the Po C implementations of IUM and demonstrates how attackers can use this type of malware to evade advanced security analysis, that is, accurate identification of target(class) victims and intention concealment in reaching non-target(class) victims. Finally, this paper provides several mitigation directions for combating IUM.

关键词： malware unbreakable property modeling mechanism accurate identification intention concealment

来源：评论

学校读者我要写书评

暂无评论

Deep Learning Compiler Optimization on Multi-Chiplet Architecture 5

Deep Learning Compiler Optimization on Multi-Chiplet Archite...

引用

5th IEEE International Conference on Artificial Intelligence Circuits and Systems, AICAS 2023

作者： Xu, Huiqing Mao, Kuang Pan, Quihong Tang, Zhaorong Wang, Mengdi Wang, Ying Zhejiang Lab Research Center for Advanced Computing Systems Hangzhou China Chinese Academy of Sciences Institute of Computing Technology Beijing China

ISBN: (纸本)9798350332674

Multi-chiplet architecture can provide a high-performance solution for new tasks such as deep learning models. In order to fully utilize chiplets and accelerate the execution of deep learning models, we present a deep learning compilation optimization framework for chiplets, and propose a scheduling method based on data dependence. Experiments show that our method improves the compilation efficiency, and the performance of the scheduling scheme is at least 1-2 times higher than the traditional algorithms. © 2023 IEEE.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

Active Voltage Balancing of Series-Connected SiC MOSFETs During Body-Diode Turn-Off Using an Active Gate Driver

引用

IEEE Transactions on Power Electronics 2025年第8期40卷 10609-10622页

作者： Rai, Ajay Ganesan, P. Hatua, Kamalesh Bhattacharya, Subhashish Department of Electrical Engineering Indian Institute of Technology Madras Chennai India Centre for Development of Advanced Computing (C-DAC) Thiruvananthapuram India Department of Electrical and Computer Engineering NC State University Raleigh NC USA

The series connection of fast-switching SiC MOSFETs presents a promising approach to enhance the efficiency, compactness, and simplicity of Medium Voltage (MV) power converters. However, managing voltage imbalances during the turn-off of MOSFETs and their body diodes introduces significant challenges. While existing literature covers passive and active voltage balancing strategies for switch turn-off (e.g., MOSFETs/IGBTs), the issue of uneven voltage distribution during body diode turn-off remains largely unexplored. This gap limits the widespread adoption of series-connected devices in MV power converters, as achieving reliable performance with minimal additional components and low switching losses remains difficult. This article presents an Active Voltage Balancing technique for handling body diode turn-off events by regulating the gate charge of parallel MOSFETs during voltage rise transients in the closedloop using Active Gate Drivers (AGD). This approach ensures uniform voltage distribution among the body diodes of seriesconnected MOSFETs. A comprehensive analysis and modelling of gate charge control during voltage transients is presented. The proposed method is experimentally validated on two seriesconnected SiC MOSFETs operating in buck mode (DC-DC) and half-bridge inverter mode (DC-AC) at a 1.5 kV DC bus voltage, with switching frequencies up to 40 kHz. © 1986-2012 IEEE.

关键词： Diodes

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：