检索结果-内蒙古大学图书馆

2024 Conference on Empirical Methods in Natural Language Processing, EMNLP 2024

作者： Lin, Jiayi Zhang, Chenyang Tong, Haibo Zhang, Dongyu Hong, Qingqing Hou, Bingxuan Wang, Junli Key Laboratory of Embedded System and Service Computing Tongji University Ministry of Education Shanghai201804 China Collaborative Innovation Center for Financial Network Security Tongji University Shanghai201804 China

ISBN: (纸本)9798891761681

Multi-Span Question Answering (MSQA) requires models to extract one or multiple answer spans from a given context to answer a question. Prior work mainly focuses on designing specific methods or applying heuristic strategies to encourage models to predict more correct predictions. However, these models are trained on gold answers and fail to consider the incorrect predictions. Through a statistical analysis, we observe that models with stronger abilities do not predict less incorrect predictions compared with other models. In this work, we propose Answering-Classifying-Correcting (ACC) framework, which employs a post-processing strategy to handle incorrect predictions. Specifically, the ACC framework first introduces a classifier to classify the predictions into three types and exclude "wrong predictions", then introduces a corrector to modify "partially correct predictions". Experiments on several MSQA datasets show that ACC framework significantly improves the Exact Match (EM) scores, and further analysis demostrates that ACC framework efficiently reduces the number of incorrect predictions, improving the quality of predictions. © 2024 Association for Computational Linguistics.

关键词： Prediction models

来源：评论

学校读者我要写书评

暂无评论

Instance-Aware Style-Swap for Disentangled Attribute-Level Image Editing 17th

Instance-Aware Style-Swap for Disentangled Attribute-Level ...

引用

17th CCF Conference on Computer Supported Cooperative Work and Social computing, ChineseCSCW 2022

作者： Zhou, Xinjiao Jiang, Bin Yang, Chao Hu, Haotian Sun, Minyu College of Computer Science and Electronic Engineering Hunan University Hunan Changsha410082 China Key Laboratory for Embedded and Network Computing of Hunan Province Hunan University Hunan Changsha410082 China

ISBN: (纸本)9789819923847

Recent studies have shown that attribute-level image editing could be achieved by modifying the latent code of style-based Generative Adversarial network (StyleGAN). However, many existing methods suffer from semantic entanglement, which leads to undesirable changes when editing a specific attribute. To solve this problem, we focus on the S space of StyleGAN2 and propose a simple and effective Optimization-based Instance-Aware Style-Swap method, which could perform a natural transformation effect on arbitrary source images and reference images by searching instance-aware swap coefficients with the guidance of prior information. By further analyzing the learned swapping coefficients, we can identify the style channels connected with a specific target attribute. We verify the effectiveness of our proposed methods in the field of face attribute editing. Extensive experiments have demonstrated that our method can achieve controllable and fine-grained image editing along various attributes. Qualitative and quantitative results show the advantages of our method compared with other semantic image editing methods. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

关键词： Generative adversarial networks

来源：评论

学校读者我要写书评

暂无评论

A novel graph kernel algorithm for improving the effect of text classification

引用

Computer Speech & Language 2026年 95卷

作者： Fan Yang Tan Zhu Jing Huang Zhilin Huang Guoqi Xie College of Computer and Information Engineering Central South University of Forestry and Technology Changsha Hunan 41004 PR China School of Computer Science and Engineering Hunan University of Science and Technology Xiangtan Hunan 411201 PR China Key Laboratory for Embedded and Network Computing of Hunan Province College of Computer Science and Electronic Engineering Hunan University Changsha Hunan 410082 PR China

Text classification is an important topic in natural language processing. In recent years, both graph kernel methods and deep learning methods have been widely employed in text classification tasks. However, previous graph kernel algorithms focused too much on the graph structure itself, such as the shortest path subgraph,while focusing limited attention to the information of the text itself. Previous deep learning methods have often resulted in substantial utilization of computational resources. Therefore,we propose a new graph kernel algorithm to address the disadvantages. First,we extract the textual information of the document using the term weighting scheme. Second,we collect the structural information on the document graph. Third, graph kernel is used for similarity measurement for text classification. We compared eight baseline methods on three experimental datasets, including traditional deep learning methods and graph-based classification methods, and tested our algorithm on multiple indicators. The experimental results demonstrate that our algorithm outperforms other baseline methods in terms of accuracy. Furthermore, it achieves a minimum reduction of 69% in memory consumption and a minimum decrease of 23% in runtime. Furthermore, as we decrease the percentage of training data, our algorithm continues to achieve superior results compared to other deep learning methods. The excellent experimental results show that our algorithm can improve the efficiency of text classification tasks and reduce the occupation of computer resources under the premise of ensuring high accuracy.

关键词： Document similarity measure Graph kernel Machine learning Term weighting Text classification

来源：评论

学校读者我要写书评

暂无评论

A recent survey on controllable text generation: A causal perspective

引用

Fundamental Research 2024年第3期5卷 1194-1203页

作者： Wang, Junli Zhang, Chenyang Zhang, Dongyu Tong, Haibo Yan, Chungang Jiang, Changjun Key Laboratory of Embedded System and Service Computing (Tongji University) Ministry of Education Shanghai 201804 China National (Province-Ministry Joint) Collaborative Innovation Center for Financial Network Security Tongji University Shanghai 201804 China

As an important subject of natural language generation, Controllable Text Generation (CTG) focuses on integrating additional constraints and controls while generating texts and has attracted a lot of attention. Existing controllable text generation approaches mainly capture the statistical association implied within training texts, but generated texts lack causality consideration. This paper intends to review recent CTG approaches from a causal perspective. Firstly, according to previous research on basic types of CTG models, it is discovered that their essence is to obtain the association, and then four kinds of challenges caused by absence of causality are introduced. Next, this paper reviews the improvements to address these challenges from four aspects, namely representation disentanglement, causal inference, knowledge enhancement and multi-aspect CTG respectively. Additionally, this paper inspects existing evaluations of CTG, especially evaluations for causality of CTG. Finally, this review discusses some future research directions for the causality improvement of CTG and makes a conclusion. © 2024

关键词： Causal inference Causality Controllable text generation Knowledge enhancement Representation disentanglement

来源：评论

学校读者我要写书评

暂无评论

TMI: Two-dimensional maintainability index for automotive software maintainability measurement

引用

Journal of Systems Architecture 2025年 166卷

作者： Jiao Xie Jialing Yang Sirong Zhao Jianmei Lei Kaiwei Fan Guoqi Xie Key Laboratory for Embedded and Network Computing of Hunan Province College of Computer Science and Electronic Engineering Hunan University Changsha 410082 Hunan China Center of Convergence of Automobile and Cyberspace Research Institute of Hunan University in Chongqing Chongqing 401120 China China Automotive Engineering Research Institute Co Ltd Electronic Communication and Software Evaluation Research Center Chongqing 401122 China

The mass adoption of autonomous driving and Advanced Driver-Assistance System (ADAS) technologies increases the complexity and maintenance cost of automotive-embedded software during model-based development. Through model-based development tools (e.g., MATLAB/Simulink), the code is automatically generated by architectural transformation; however, unreasonable software architecture design could make the final generated code be difficult to maintain. Currently, both Microsoft Maintainability Index (MMI) and Bosch Maintainability Index (BMI) are used to measure the maintainability of code, but neither software architecture. We establish a four-layer maintainability index model by analyzing the characteristics of automotive software architecture, software quality model, and industry-related standards. We then propose a Two-dimensional Maintainability Index (TMI), meaning the code and architecture dimensions are simultaneously considered. This is the first work of maintainability measurement to combine code and architecture dimensions while ensuring low redundancy. TMI utilizes nine index metrics to guide architecture refactoring and code optimization, significantly improving software maintainability.

关键词： MATLAB

来源：评论

学校读者我要写书评

暂无评论

New binary artificial bee colony for the 0-1 knapsack problem 9th

New binary artificial bee colony for the 0-1 knapsack proble...

引用

9th International Conference on Swarm Intelligence, ICSI 2018

作者： Nouioua, Mourad Li, Zhiyong Jiang, Shilong The College of Computer Science and Electronic Engineering Hunan University Changsha China Key Laboratory for Embedded and Network Computing of Hunan Province Changsha China PKU-HKUST Shenzhen-HongKong Institution Shenzhen China

ISBN: (纸本)9783319938141

The knapsack problem is one of the well known NP-Hard optimization problems. Because of its appearance as a sub-problem in many real world problems, it attracts the attention of many researchers on swarm intelligence and evolutionary computation community. In this paper, a new binary artificial bee colony called NB-ABC is proposed to solve the 0-1 knapsack problem. Instead of the search operators of the original ABC, new binary search operators are designed for the different phases of the ABC algorithm, namely the employed, the onlooker and the scout bee phases. Moreover, a novel hybrid repair operator called (HRO) is proposed to repair and improve the infeasible solutions. To assess the performance of the proposed algorithm, NB-ABC is compared with two other existing algorithms, namely GB-ABC and BABC-DE, for solving the 0-1 knapsack problem. Based on a set of 15 0-1 high dimensional knapsack problems classified in three categories. the experimental results in view of many criteria show the efficiency and the robustness of the proposed NB-ABC. © Springer International Publishing AG, part of Springer Nature 2018.

关键词： Repair

来源：评论

学校读者我要写书评

暂无评论

Sequential channel selection for decentralized cognitive radio sensor network based on modified Q-Learning algorithm

Sequential channel selection for decentralized cognitive rad...

引用

International Conference on Fuzzy Systems and Knowledge Discovery (FSKD)

作者： Fanzi Zeng Hanshan Liu Jisheng Xu Key Laboratory for Embedded and Network Computing of Hunan Province Hunan University Changsha China

Cognitive radio sensor network (CRSN) is a novel wireless sensor network, which can effectively alleviate the scarcity of spectrum resource by equipping the cognitive technology. Usually, CRSN deploys numerous sensor nodes around the monitoring area. But the energy and storage of those nodes are both limited. Compared with the centralized CRSN, we focus more on a decentralized situation without the control center or cluster in this paper, especially on a circumstance with few information exchanges among sensor nodes. In the case of that, our goal is to help each sensor node to find an optimal strategy of channel selection in the decentralized CRSN quickly, and then, we apply a Multi-Agent Best-Response Q-Learning algorithm based on Timesharing-Tracking Framework to sequential channel selection for the first time. The algorithm not only has a low storage cost, but also can overcome the poor convergence of strategy caused by Multi-Agent simultaneous learning. For comparison, the optimal channel assignment in centralized network, the Multi-Agent Independent Q-Learning and the Win-or-Learn-Fast Policy Hill Climbing these three algorithms are taken into account, and experimental data validates that our proposed algorithm is feasible in decentralized network.

关键词： Sensors Throughput Wireless sensor networks Convergence Algorithm design and analysis Cognitive radio Approximation algorithms

来源：评论

学校读者我要写书评

暂无评论

A novel energy efficient cooperative spectrum sensing scheme for cognitive radio sensor network based on evolutionary game

A novel energy efficient cooperative spectrum sensing scheme...

引用

IEEE Workshop on Local and Metropolitan Area networks (LANMAN)

作者： Xiaolong Ma Fanzi Zeng Jisheng Xu Key Laboratory for Embedded and Network Computing of Hunan Province Hunan University Changsha China

ISBN: (纸本)9781467367639

In this paper, we focused on how to get CRs collaborate effectively as well as maximize energy savings in cognitive radio sensor network. We have established a model of evolutionary game between a cognitive sensors. In order to let cognitive sensors sense spectrum effectively, we designed a contribution-punishment mechanism, which can stimulate high SNR sensors to participate in spectrum sensing. Additionally, in order to reduce the wasted energy of sensors in idle, we introduced periodic sleep-listen mechanism. Finally, we simulated behavior dynamics for a 6-player polymorphic population game and compared our proposed scheme with non-periodic sleep-listen mechanism scheme. Result shows that our proposed scheme can stimulate high SNR sensors to participate in spectrum sensing effectively and can make the system more energy efficient.

关键词： Sensors Signal to noise ratio Games Sociology Statistics Energy consumption Energy efficiency

来源：评论

学校读者我要写书评

暂无评论

A High-performance DAG Task Scheduling Algorithm for Heterogeneous networked embedded Systems

A High-performance DAG Task Scheduling Algorithm for Heterog...

引用

IEEE International Conference on Advanced Information networking and Applications

作者： Guoqi Xie Renfa Li Xiongren Xiao Yuekun Chen Key Laboratory for Embedded and Network Computing of Hunan Province Hunan University Changsha China

ISBN: (纸本)9781479936311

A high-performance scheduling for a DAG (Directed Acyclic Graph) task graph on heterogeneous networked embedded systems or parallel and distributed systems is to maximize concurrency and minimize inter-processor communication. Most of the algorithms using upward rank value for task prioritizing and earliest finish time for processor assignment. But both approaches ignored the heterogeneity of system and could not create accurate and efficient schedules. Yet no one has doubled about and recognized that. A fully heterogeneous task scheduling algorithm is proposed to address the above problems in this paper. The fundamentals of DAG model and corresponding algorithms are investigated. New concepts called Heterogeneous Upward Rank Value (HURV) and Heterogeneous Priority Rank Value (HPRV) are defined. An algorithm called Heterogeneous Select Value (HSV) is proposed in paper. Both benchmark and extensive experimental evaluation demonstrate the significant improvements in proposed algorithm.

关键词： heterogeneous networked embedded systems heterogeneous upward rank value heterogeneous select value DAG

来源：评论

学校读者我要写书评

暂无评论

A Power Model Combined of Architectural Level and Gate Level for Multicore Processors

A Power Model Combined of Architectural Level and Gate Level...

引用

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

作者： Manman Peng Yan Hu Key Laboratory for Embedded and Network Computing of Hunan Province Hunan University Changsha China

ISBN: (纸本)9781479914449

Low power consumption is becoming a critical factor for multicore processors. As the multicore processor design complexity increases, power estimation for multicore processors has gained more importance. This paper presents a new power model combined of architectural level and gate level for multicore processors. The model maps the multicore processors to a combination of building blocks, and estimates the gate-level power of these blocks using parameterized RTL. Then, the power numbers are made in the form of look-up tables, and integrated in architecture simulators. The experiments show that for peak power estimation, an excellent accuracy has been reached and simulation performance is greatly improved compared to the gate level.

关键词： Program processors Computational modeling Multicore processing Logic gates Random access memory Hardware Analytical models

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：