检索结果-内蒙古大学图书馆

12th International Conference on Learning Representations, ICLR 2024

作者： Feng, Zijian Zhou, Hanzhang Zhu, Zixiao Qian, Junlang Mao, Kezhi Institute of Catastrophe Risk Management Interdisciplinary Graduate Programme Nanyang Technological University Singapore School of Electrical and Electronic Engineering Nanyang Technological University Singapore Future Resilient Systems Programme Singapore-ETH Centre CREATE Campus Singapore

Prompts play a crucial role in guiding the responses of Large Language Models (LLMs). However, the intricate role of individual tokens in prompts, known as input saliency, in shaping the responses remains largely underexplored. Existing saliency methods either misalign with LLM generation objectives or rely heavily on linearity assumptions, leading to potential inaccuracies. To address this, we propose Token Distribution Dynamics (TDD), a simple yet effective approach to unveil and manipulate the role of prompts in generating LLM outputs. TDD leverages the robust interpreting capabilities of the language model head (LM head) to assess input saliency. It projects input tokens into the embedding space and then estimates their significance based on distribution dynamics over the vocabulary. We introduce three TDD variants: forward, backward, and bidirectional, each offering unique insights into token relevance. Extensive experiments reveal that the TDD surpasses state-of-the-art baselines with a big margin in elucidating the causal relationships between prompts and LLM outputs. Beyond mere interpretation, we apply TDD to two prompt manipulation tasks for controlled text generation: zero-shot toxic language suppression and sentiment steering. Empirical results underscore TDD's proficiency in identifying both toxic and sentimental cues in prompts, subsequently mitigating toxicity or modulating sentiment in the generated content. © 2024 12th International Conference on Learning Representations, ICLR 2024. All rights reserved.

关键词： Zero-shot learning

来源：评论

学校读者我要写书评

暂无评论

Cloud-Honeypot and Differential Privacy Empowered Interactive Heat Protection against Public Opinion Analysis 8

Cloud-Honeypot and Differential Privacy Empowered Interactiv...

引用

8th IEEE International Conference on Smart Cloud, SmartCloud 2023

作者： Sun, Zhengyi Pan, Qianqian Li, Zhaohui Wu, Jun Graduate School of Information Production and Systems Waseda University Kitakyushu Japan School of Engineering The University of Tokyo Tokyo Japan School of Maritime Economics and Management Dalian Maritime University Dalian China

ISBN: (纸本)9798350313505

With the rapid development of cloud social networks, the expression of opinions by individuals has witnessed significant growth. However, unregulated dissemination of public opinion can lead to severe consequences. This paper addresses the challenges associated with public opinion heat analysis, focusing on the potential misuse of user data by malicious entities. We propose a privacy protection mechanism that combines differential privacy theory and cloud-honeypot technologies, which is against scraper to ensure the legitimate use of data and safeguard user privacy. We propose a differential privacy mechanism with Laplace noise, which adds Laplace noise to the number of reposts, comments, and likes of the Weibo posts published by users, making it impossible for attackers to calculate the correct interactive heat and thus protecting the privacy of Weibo interaction popularity. At the same time, we propose a privacy protection mechanism which is applicable in combination with honeypot technology. After the cloud server identifies that the visitor is a malicious scraper, it redirects them to the honeypot webpage. This modified webpage incorporates additional Laplace noise, thereby enhancing the security measure. Additionally, we design an experiment to simulate the impact of adding noise to the data and compare the results to verify the feasibility of the differential privacy mechanism in countering scrapers. Finally, we discuss future directions for anti-scraping strategies based on differential privacy. ©2023 IEEE.

关键词： Network security

来源：评论

学校读者我要写书评

暂无评论

Preferences in Construction Workers Selection to a Project Team: A Conjoint Analysis Approach 24

Preferences in Construction Workers Selection to a Project T...

引用

6th International Conference on management Science and Industrial engineering, MSIE 2024

作者： Bonete, Lorenz Ian B. Prasetyo, Yogi Tri Ahmed, Nagib Ismail Cahigas, Maela Madel L. Nadlifatin, Reny Ayuwati, Irene Dyah School of Industrial Engineering and Engineering Management Mapúa University Manila Philippines School of Graduate Studies Mapúa University Manila Philippines International Bachelor Program in Engineering Yuan Ze University Chung-Li Taiwan Department of Industrial Engineering and Management Yuan Ze University Chung-Li Taiwan Department of Information Systems Institut Teknologi Sepuluh Nopember Kampus ITS Sukolilo Surabaya60111 Indonesia University of Surabaya Surabaya60284 Indonesia

ISBN: (纸本)9798400717130

Construction teams are responsible for building complex projects within the allotted budget, time and expected quality. Thus, team members must be able to work with each other to avoid any complications during the construction process. As such, team supervisors must select from a pool of applicants or from their current roster of workers, which would be composed of different personalities and capabilities. This paper aims to identify the preferences of team supervisors when building up their team using conjoin analysis. The study found out that supervisors would like to have workers that respects and values their work and can work harmoniously with others. This study is beneficial for the human resource department of construction companies when recruiting workers, for the management when trying to develop their current personnel and for the workers who are trying to improve themselves. © 2024 ACM.

关键词： Supervisory personnel

来源：评论

学校读者我要写书评

暂无评论

Factors that Influence E-Satisfaction and Their Impact on Online Purchase Intention of Telemedicine Applications 9

Factors that Influence E-Satisfaction and Their Impact on On...

引用

9th IEEE International Conference on Computing, engineering and Design, ICCED 2023

作者： Gui, Anderes So, Idris Gautama Novantia, Dea Princes, Elfindah Sianipar, Nesti Fronika Haron, Hasnah Bina Nusantara University Information System Program School of Information Systems Jakarta Indonesia Bina Nusantara University Binus Business School Management Department Jakarta Indonesia Bina Nusantara University Binus Graduate Program-Master of Information Systems Management Information Systems Management Department Jakarta Indonesia Bina Nusantara University Faculty of Engineering Food Technology Department Jakarta Indonesia Universiti Sains Islam Malaysia Faculty of Economics and Muamalat Nilai Malaysia

ISBN: (纸本)9798350370126

This study embarked on a rigorous examination of the factors driving user satisfaction and usage behavior in the context of telehealth applications. Utilizing a well- structured Google Form survey, distributed through an array of social media platforms, including WhatsApp, Discord, Instagram, and Line, the research gathered 251 responses. The purposive sampling technique was employed to ensure that the data captured the experiences of individuals who had actively engaged with telehealth services. One hypothesis, relating to the influence of Online Reviews, was found to be unsupported. This surprising result suggests that, contrary to expectations, the opinions and reviews shared online do not significantly affect users' satisfaction or their behavioral decisions in the context of telehealth applications. This finding highlights the need for a more nuanced understanding of the factors influencing user preferences and choices in this rapidly evolving sector. Conversely, other hypotheses examining factors such as Social Influence, Facilitating Conditions, Perceived Reliability, Price Value, and Purchase Intention were substantiated. The positive influence of these factors on customer usage behavior underscores their significance in shaping user experiences and satisfaction within the telehealth ecosystem. Furthermore, the study posits that facilitating conditions should be closely aligned with technological advancements, as this alignment is conducive to the development of enhanced application features. This insight suggests that telehealth providers should continuously innovate and integrate advanced technologies to ensure that the facilitating conditions meet the ever-evolving needs and expectations of users. © 2023 IEEE.

关键词： Sales

来源：评论

学校读者我要写书评

暂无评论

Mean-variance and Safety-first Portfolio Selection Strategy for High-volume/High-value Traded Stocks in Philippine Stock Exchange (PSE)

Mean-variance and Safety-first Portfolio Selection Strategy ...

引用

2024 IEEE International Conference on Industrial engineering and engineering management, IEEM 2024

作者： De Guzman, T.C. Young, M.N. Prasetyo, Y.T. Persada, S.F. Nadilfatin, R. School of Graduate Studies Mapua University Manila Philippines School of Industrial Engineering and Engineering Management Mapua University Manila Philippines Department of Industrial Engineering and Management Yuan Ze University Taoyuan City Taiwan International Bachelor Program in Engineering Yuan Ze University Taoyuan City Taiwan Entrepreneurship Business Creation Business School Binus University West Jakarta Indonesia Department of Information Systems Institut Teknologi Sepuluh Nopember Surabaya Indonesia

ISBN: (纸本)9798350386097

This research evaluates the effectiveness of Mean-Variance (MV) and Safety-First (SF) portfolio selection strategies for high-volume/high-value traded stocks on the Philippine Stock Exchange (PSE). Analyzing data from 288 stocks over 7,670 trading days (30 years), we estimated stock returns, assigned probability weights, and conducted optimization models. The results show that the Mean-Variance (MV0.5) strategy achieved the best balance between risk and return, boasting the highest mean and cumulative returns with moderate volatility. The Safety-First (SF) strategy, however, yielded negative cumulative returns, indicating potential issues. The Market strategy had the lowest mean return but also the least risk, appealing to risk-averse investors. The Portfolio Mean-Variance (PMV) strategy provided a reasonable balance of returns and risk. Statistical analysis revealed that only the MV0.5 strategy outperformed market returns. These findings offer crucial insights into risk assessment and portfolio management for investors evaluating these strategies against market performance. © 2024 IEEE.

关键词： Marketplaces

来源：评论

学校读者我要写书评

暂无评论

Integration of human factors, cognitive ergonomics, and artificial intelligence in the human-machine interface for additive manufacturing

Integration of human factors, cognitive ergonomics, and arti...

引用

作者： Doran, Eric Bommer, Sharon Badiru, Adedeji Department of Engineering Management Systems and Technology School of Engineering University of Dayton DaytonOH45469 United States Graduate School of Engineering & Management Air Force Institute of Technology Wright-Patterson Air Force Base OH45433 United States

As additive manufacturing transitions from manufacturing prototypes to rapid manufacturing, more human factors considerations must be assessed and integrated for improved work design. This review paper provides an overview of human-machine integration for human factors, cognitive ergonomics, and artificial intelligence to improve the performance output of the additive manufacturing process. In addition, case studies are shared to provide integration concepts for artificially intelligent systems. It is anticipated that the contents of this review paper will pave the path for further research into the integration of human factors and cognitive ergonomics for artificial intelligence in additive manufacturing. Copyright © 2022 Inderscience Enterprises Ltd.

关键词： Integration

来源：评论

学校读者我要写书评

暂无评论

A Private Blockchain System based on Zero Trust Architecture 26

A Private Blockchain System based on Zero Trust Architecture

引用

26th International Conference on Advanced Communications Technology, ICACT 2024

作者： Chang, Yao-Chung Lin, Yu-Shan Sangaiahc, Arun Kumar Wu, Hsin-Te National Taitung University Department of Computer Science and Information Engineering Taitung Taiwan National Taitung University Department of Information Science and Management Systems Taitung Taiwan National Yunlin University of Science and Technology Department of International Graduate School of Artificial Intelligence Yunlin Taiwan

ISBN: (纸本)9791188428120

During the outbreak of COVID-19, many enterprises massively created Virtual Private Networks (VPNs) for companies to cooperate;however, these accounts lacked efficient management after the epidemic, leading to data leakage or the suffering of malicious attacks. Consequently, several firms have started to build private blockchains for data preservation and verification. Private blockchains are usually built internally in companies;once a firm's internal network opens for massive external logins, many servers and private blockchains cannot work properly to protect data. In general, private blockchains store critical personal information or relevant confidential data;once a private blockchain opens to external access, all the information in the network nodes will be exposed, making private them lose their protection functions. This paper proposes a security mechanism using a private blockchain system based on zero trust architecture. The zero trust architecture tracks every user's network conditions and analyses whether their behaviors are authorized. Additionally, the system utilizes micro-segmentation to divide the private blockchain, preventing the system from malicious attacks. The proposed system employs user multi-factor authentication to identify users, and the zero trust architecture tracks and analyzes if users' behaviors are reasonable. This method effectively ensures corporate networks' security and enables private blockchain to filter legal and authorized users to access and verify. © 2024 Global IT Research Institute - GIRI.

关键词： Virtual private networks

来源：评论

学校读者我要写书评

暂无评论

Portfolio Selection Using Mean Variance Theory Based on the 30-year Historical Returns of Selected Subsectors of the Philippine Industrial Sector

Portfolio Selection Using Mean Variance Theory Based on the ...

引用

2024 IEEE International Conference on Industrial engineering and engineering management, IEEM 2024

作者： Yamanaka, A.D. Young, M.N. Chuahay, T.N. Pratseyo, Y. Persada, S. Nadlifatin, R. School of Graduate Studies Mapua University Manila Philippines School of Industrial Engineering and Engineering Management Mapua University Manila Philippines Department of Finance Chung Yuan University Taoyuan City Taiwan Department of Industrial Engineering and Management Yuan Ze University Taoyuan City Taiwan International Bachelor Program in Engineering Yuan Ze University Taoyuan City Taiwan Entrepreneurship Business Creation Business School Binus University West Jakarta Indonesia Department of Information Systems Institut Teknologi Sepuluh Nopember Surabaya Indonesia

ISBN: (纸本)9798350386097

Portfolio selection is an important part of investment management that aims to select the right combination of assets to achieve an optimal risk-return ratio. This study integrates the Mean-Variance Theory utilizing the 30-year historical returns of the Philippine Stock Exchange. Three subsectors of the Industrial Sector were used as the investment pool. The back-test results showed that the Mean-Variance portfolios of selected subsectors can outperform the industrial sector and not the market. Stakeholders must be cautious when investing in these three subsectors, considering the values obtained from the simulations. There are recommended portfolio selections wherein companies have a specific allocation that investors and financial managers can use whenever they decide to invest in these subsectors. Overall, this study offers a framework for investors to use as a guide in diversifying their stakes in the Philippine Stock Exchange. © 2024 IEEE.

关键词： Decentralized finance

来源：评论

学校读者我要写书评

暂无评论

Automatic Gait Classification using Arm Acceleration based on Stature and Gait Speed 4

Automatic Gait Classification using Arm Acceleration based o...

引用

4th IEEE Global Conference on Life Sciences and Technologies, LifeTech 2022

作者： Kitagawa, Kodai Wada, Chikamune Toya, Nobuyuki Kyushu Institute of Technology Graduate School of Life Science and Systems Engineering Kitakyushu Japan Hokkaido Information University Department of Medical Management and Informatics Ebetsu Japan

ISBN: (纸本)9781665419048

Unsuitable foot clearance and step length lead falling accidents;thus, these gait parameters should be monitored for fall prevention. According to previous study, users prefer wrist as a position to wear wearable device for daily gait monitoring. Therefore, we have been developing gait monitoring method using wearable device on a wrist. Our previous method classifies gaits with different foot clearance or step length by machine learning technique using 3-axis acceleration data obtained from wearable device on a wrist. The results of our previous study showed that this method could classify gaits for each step. However, accuracy of our previous method was insufficient because individual difference of arm acceleration due to stature or gait speed were not considered. The objective of this study is to propose the gait classification method based on individual difference. The proposed method classifies three gaits with different foot clearance and step length by machine learning technique using combination of arm acceleration, stature, and gait speed. The proposed method was tested for 1800 steps performed from ten participants. The proposed method could classify three gaits with accuracy greater than 0.9. These results indicate possibility that the proposed method can be used for gait monitoring using wearable device. © 2022 IEEE.

关键词： Acceleration

来源：评论

学校读者我要写书评

暂无评论

Ideal Secret Sharing Schemes: Combinatorial Characterizations, Certain Access Structures, and Related Geometric Problems

arXiv

引用

arXiv 2023年

作者： Fuji-Hara, Ryoh Miao, Ying Department of Social Systems and Management Graduate School of Systems and Information Engineering University of Tsukuba Tsukuba305-8573 Japan

An ideal secret sharing scheme is a method of sharing a secret key in some key space among a finite set of participants in such a way that only the authorized subsets of participants can reconstruct the secret key from their shares which are of the same length as that of the secret key. The set of all authorized subsets of participants is the access structure of the secret sharing scheme. In this paper, we derive several properties and restate the combinatorial characterization of an ideal secret sharing scheme in Brickell-Stinson model in terms of orthogonality of its representative array. We propose two practical models, namely the parallel and hierarchical models, for access structures, and then, by the restated characterization, we discuss sufficient conditions on finite geometries for ideal secret sharing schemes to realize these access structure models. Several series of ideal secret sharing schemes realizing special parallel or hierarchical access structure model are constructed from finite projective planes. Copyright © 2023, The Authors. All rights reserved.

关键词： Geometry

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：