检索结果-内蒙古大学图书馆

arXiv 2024年

作者： Zhong, Qihuang Ding, Liang Liu, Juhua Du, Bo Tao, Dacheng School of Computer Science National Engineering Research Center for Multimedia Software Institute of Artificial Intelligence Hubei Key Laboratory of Multimedia and Network Communication Engineering Wuhan University China The University of Sydney Australia Nanyang Technological University Singapore

With the development of instruction-tuned large language models (LLMs), improving the safety of LLMs has become more critical. However, the current approaches for aligning the LLMs output with expected safety usually require substantial training efforts, e.g., high-quality safety data and expensive computational resources, which are costly and inefficient. To this end, we present Reverse prOmpt contraStive dEcoding (ROSE), a simple-yet-effective method to directly boost the safety of existing instruction-tuned LLMs without any additional training. The principle of ROSE is to improve the probability of desired safe output via suppressing the undesired output induced by the carefully-designed reverse prompts. Experiments on 6 safety and 2 general-purpose tasks show that, our ROSE not only brings consistent and significant safety improvements (up to +13.8% safety score) upon 5 types of instruction-tuned LLMs, but also benefits the general-purpose ability of LLMs. In-depth analyses explore the underlying mechanism of ROSE, and reveal when and where to use it. Copyright © 2024, The Authors. All rights reserved.

关键词： Computational linguistics

来源：评论

学校读者我要写书评

暂无评论

Intent-aware Prompt Learning for Medical Question Summarization

Intent-aware Prompt Learning for Medical Question Summarizat...

引用

IEEE International Conference on Bioinformatics and Biomedicine (BIBM)

作者： Leilei Zhang Junfei Liu School of Software & Microelectronics Peking University Beijing China School of Electronics Engineering and Computer Science Peking University Beijing China

ISBN: (数字)9781665468190

ISBN: (纸本)9781665468206

Consumer health question summarization aims to generate concise questions expressing the minimum information needed to find the correct answer to the original user queries, which can significantly improve the success rate of retrieving relevant answers in automated medical question answering systems. Recently, prompt learning has emerged as a new paradigm of leveraging pre-trained language models (PLMs) with promising results in downstream tasks. Existing prompt-based methods assume that all samples in a task share the same prompt. However, the health questions raised by consumers in search engines focus on different types of medical entities and cover different purposes. It is not sufficient to simply use the same prompt regardless of consumer intent awareness, which likely leads to meaningless or substandard quality summaries. In this paper, we propose an intent-aware prompt learning (InPL) method, which consists of a question intent recognition module, a dynamic prompt module and a medical summarization generation module to model the medical summarization task in a unified transformer. Our InPL architecture generates high-quality summaries capturing more consumer health intent and encourages machines to generate more factually correct summaries. Extensive experiments on the popular medical summarization datasets demonstrate the superiority of our approach.

关键词： Biological system modeling Aggregates Search engines Transformers Question answering (information retrieval) Generators Task analysis

来源：评论

学校读者我要写书评

暂无评论

Optimizing Mobile-Friendly Viewport Prediction for Live 360-Degree Video Streaming

引用

IEEE Transactions on Mobile Computing 2025年

作者： Zhang, Lei Chen, Peng Zhang, Cong Pan, Cheng Long, Tao Xu, Weizhen Cui, Laizhong Liu, Jiangchuan Shenzhen University College of Computer Science and Software Engineering Guangdong Shenzhen518060 China Guangdong Shenzhen518000 China University of Hong Kong Department of Electrical and Electronic Engineering Pok Fu Lam Hong Kong Simon Fraser University School of Computing Science BurnabyBCV5A 1S6 Canada

Viewport prediction is the crucial task for adaptive 360-degree video streaming, as the bitrate control algorithms usually require the knowledge of the user's viewing portions of the frames. Various methods are studied and adopted for viewport prediction from less accurate statistic tools to highly calibrated deep neural networks. Conventionally, it is difficult to implement sophisticated deep learning methods on mobile devices, which have limited computation capability. In this work, we propose an advanced learning-based viewport prediction approach and carefully design it to minimize transmission and computation overhead for mobile terminals. To improve viewport prediction accuracy, we utilize both spatial information through a saliency prediction model and temporal information through a modified LSTM model. Different computations introduced by the neural network models are distributed across the network to keep the computation light on mobile devices. To better adapt to the content dynamics in live streaming, we employ the model-agnostic meta-learning (MAML) method for video saliency prediction. The learned saliency prediction model with optimized initialization via offline meta-training can be fast fine-tuned online using a few samples. We further discuss how to integrate this mobile-friendly viewport prediction (MFVP) approach into a typical 360-degree video live streaming system by formulating and solving the bitrate adaptation problem. Extensive experiment results demonstrate that our approach achieves real-time prediction for live video streaming and surpasses existing methods in prediction accuracy on mobile terminals, which, together with our bitrate adaptation algorithm, significantly improves the streaming QoE from various aspects. Compared to baseline methods, MFVP achieves a 4.7–28.7% improvement in accuracy and demonstrates faster adaptability to dynamic content changes, enabling rapid fine-tuning and adjustment. When integrated into a streaming syste

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

A Deep Reinforcement Learning Perspective on Adaptive Federated Dropout

A Deep Reinforcement Learning Perspective on Adaptive Federa...

引用

IEEE Internatonal Conference on Mobile Adhoc and Sensor Systems (MASS)

作者： Tonghui Sun Yan Huang Zhenzhen Xie Junjie Pang Xingyun Chen Zhipeng Cai College of Computer Science and Technology Qingdao University Qingdao China College of Computing and Software Engineering Kennesaw State University Atlanta GA USA School of Computer Science and Technology Shandong University Qingdao China Department of Computer Science Georgia State University Atlanta GA USA

Federated learning is a distributed learning solution that achieves high-quality machine learning models while ensuring privacy and collaboration among various end devices. However, different kinds of end devices can lead to an unstable training process due to limited and dynamic communication and computation resources. Federated Dropout is an important technique for mitigating such resource bottlenecks by randomly removing a fixed percentage of activations of model components. However, most of the existing FL dropout methods cannot effectively utilize the distinguished characteristics of different clients or fail to automatically adapt to end device heterogeneity. In this paper, we thus propose Adaptive Federated Dropout with Reinforcement Learning (AFD-RL). AFD-RL employs reinforcement learning technology to adapt to the clients’ unbalanced computation and communication resources and provide a personalized dropout strategy for each client. By automatically selecting suitable model components for each client in each communication round, AFD-RL achieves better accuracy with fewer communication rounds. Extensive experiments validate the effectiveness of the proposed approach in improving training efficiency and inference accuracy.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Detection of Alzheimer's disease using deep learning models: A systematic literature review

引用

Informatics in Medicine Unlocked 2024年 50卷

作者： Mohammed, Eqtidar M. Fakhrudeen, Ahmed M. Alani, Omar Younis Department of Computer Science College of Computer Science and Information Technology University of Kirkuk Kirkuk Iraq Department of Software College of Computer Science and Information Technology University of Kirkuk Kirkuk Iraq School of Science Engineering & Environmnet University of Salford Manchester M5 4WT United Kingdom

Alzheimer's disease (AD) is a progressive neurological disease considered the most common form of late-stage dementia. Usually, AD leads to a reduction in brain volume, impacting various functions. This article comprehensively analyzes the AD context in fivefold main topics. Firstly, it reviews the main imaging techniques used in diagnosing AD disease. Secondly, it explores the most proposed deep learning (DL) algorithms for detecting the disease. Thirdly, the article investigates the commonly used datasets to develop DL techniques. Fourthly, we conducted a systematic review and selected 45 papers published in highly ranked publishers (science Direct, IEEE, Springer, and MDPI). We analyzed them thoroughly by delving into the stages of AD diagnosis and emphasizing the role of preprocessing techniques. Lastly, the paper addresses the remaining practical implications and challenges in the AD context. Building on the analysis, this survey contributes to covering several aspects related to AD disease that have not been studied thoroughly. © 2024

关键词： AD Alzheimer's disease Challenges Datasets Deep learning Mild cognitive impairment

来源：评论

学校读者我要写书评

暂无评论

Curriculum Contrastive Learning for COVID-19 FAQ Retrieval

Curriculum Contrastive Learning for COVID-19 FAQ Retrieval

引用

IEEE International Conference on Bioinformatics and Biomedicine (BIBM)

作者： Leilei Zhang Junfei Liu School of Software & Microelectronics Peking University Beijing China School of Electronics Engineering and Computer Science Peking University Beijing China

ISBN: (数字)9781665468190

ISBN: (纸本)9781665468206

Medical Frequently Asked Question (FAQ) retrieval aims to find the most relevant question-answer pairs for a given user query, which is of great significance for enhancing people medical health awareness and knowledge. However, due to medical data privacy and labor-intensive labeling, there is a lack of large-scale question-matching training datasets. Previous methods directly use the collected question-answer pairs on search engines to train retrieval models, which achieved poor performance. Inspired by recent advances in contrastive learning, we propose a novel contrastive curriculum learning framework for modeling user medical queries. First, we design different data augmentation methods to generate positive samples and different types of negative samples. Second, we propose a curriculum learning strategy that associates difficulty levels with negative samples. Through a contrastive learning process from easy to hard, our method achieves excellent results on two medical datasets.

关键词： Training COVID-19 Data privacy Search engines Data models Labeling Bioinformatics

来源：评论

学校读者我要写书评

暂无评论

An Introduction to Real-Time Object-Z

引用

Formal Aspects of Computing 2022年第2期13卷 128–141页

作者： Graeme Smith Ian Hayes Software Verification Research Centre University of Queensland Brisbane Australia AU School of Computer Science and Electrical Engineering University of Queensland Brisbane Australia AU

This paper presents Real-Time Object-Z: an integration of the object-oriented, state-based specification language Object-Z with the timed trace notation of the timed refinement calculus. This integration provides a method of formally specifying and refining systems involving continuous variables and real-time constraints. The basis of the integration is a mapping of the existing Object-Z history semantics to timed traces.

关键词： Keywords: Object-Z Real-time refinement Real-time specification Timed refinement calculus

来源：评论

学校读者我要写书评

暂无评论

Transfer learning for accurate brain tumor classification in MRI: a step forward in medical diagnostics

引用

Discover oncology 2025年第1期16卷 1040页

作者： Muhammad Adnan Khan Muhammad Zahid Hussain Shahid Mehmood Muhammad Farhan Khan Munir Ahmad Tehseen Mazhar Tariq Shahzad Mamoon M Saeed Department of Software Faculty of Artificial Intelligence and Software Gachon University Seongnam-si Gyeonggido 13120 Republic of Korea. Faculty of Information Technology and Computer Science University of Central Punjab Lahore 54000 Pakistan. Department of Computer Science Bahria University Lahore Campus Lahore 54000 Pakistan. Department of Forensic Sciences University of Health Sciences Lahore 54000 Pakistan. University College Korea University Seoul 02841 Republic of Korea. School of Computer Science National College of Business Administration and Economics Lahore 54000 Pakistan. tehseenmazhar719@***. Department of Computer Science School Education Department Government of Punjab Layyah 31200 Pakistan. tehseenmazhar719@***. Department of Computer Engineering COMSATS University Islamabad Sahiwal Campus Sahiwal 57000 Pakistan. Department of Communications and Electronics Engineering Faculty of Engineering University of Modern Sciences (UMS) 00967 Sana Yemen. dr.mamoon@***.

Brain tumor classification is critical for therapeutic applications that benefit from computer-aided diagnostics. Misdiagnosing a brain tumor can significantly reduce a patient's chances of survival, as it may lead to ineffective treatments. This study proposes a novel approach for classifying brain tumors in MRI images using Transfer Learning (TL) with state-of-the-art deep learning models: AlexNet, MobileNetV2, and GoogleNet. Unlike previous studies that often focus on a single model, our work comprehensively compares these architectures, fine-tuned specifically for brain tumor classification. We utilize a publicly available dataset of 4,517 MRI scans, consisting of three prevalent types of brain tumors-glioma (1,129 images), meningioma (1,134 images), and pituitary tumors (1,138 images)-as well as 1,116 images of normal brains (no tumor). Our approach addresses key research gaps, including class imbalance, through data augmentation and model efficiency, leveraging lightweight architectures like MobileNetV2. The GoogleNet model achieves the highest classification accuracy of 99.2%, outperforming previous studies using the same dataset. This demonstrates the potential of our approach to assist physicians in making rapid and precise decisions, thereby improving patient outcomes. The results highlight the effectiveness of TL in medical diagnostics and its potential for real-world clinical deployment. This study advances the field of brain tumor classification and provides a robust framework for future research in medical image analysis.

关键词： AlexNet Brain tumor classification Deep learning GoogleNet MobileNetv2 Transfer learning

来源：评论

学校读者我要写书评

暂无评论

Global Decision Making Support for Complex System Development

Global Decision Making Support for Complex System Developmen...

引用

IEEE International Conference on Requirements engineering

作者： Lola Burgueño Damien Foures Benoit Combemale Jörg Kienzle Gunter Mussbacher ITIS Software University of Malaga Malaga Spain DDMS Airbus Group Toulouse France ESIR & IRISA University of Rennes Rennes France School of Computer Science McGill University Montreal Canada Department of Electrical and Computer Engineering McGill University Montreal Canada Department of Electrical and Computer Engineering INRIA Rennes France

ISBN: (数字)9798350395112

ISBN: (纸本)9798350395129

To succeed with the development of modern and complex systems (e.g., aircrafts or production systems), organizations must have the agility to adapt faster to constantly evolving requirements in order to deliver more reliable and optimized solutions that can be adapted to the needs and environments of their stakeholders including users, customers, suppliers, and partners. However, stakeholders do not have sufficiently explicit and systematic support for global decision making, considering the vast decision space and complex inter-relationships. This decision space is characterized by increasing yet inadequately represented variability and the uncertainty of the impact of decisions on stakeholders and the solution space. This leads to an ad-hoc decision making process that is slow, error-prone, and often favors local knowledge over global, organization-wide objectives. As a result, one team's design decisions may impose too restrictive requirements on another team. In this paper, we evaluate our understanding of global decision making in the context of complex system development based on a conceptual model which explicitly represents and manages decision spaces including variability and impacts. We have conducted our evaluation by means of an exploratory case study where we interviewed domain experts with an average of 20 years of experience in complex system industries and report the key findings and remaining challenges. In the future, we aim at providing explicit and systematic tool-supported approaches for global decision making support for complex systems.

关键词： Industries Production systems Systematics Uncertainty Decision making Organizations Stakeholders

来源：评论

学校读者我要写书评

暂无评论

Financial Crimes in Web3-empowered Metaverse: Taxonomy, Countermeasures, and Opportunities

arXiv

引用

arXiv 2022年

作者： Wu, Jiajing Lin, Kaixin Lin, Dan Zheng, Ziye Huang, Huawei Zheng, Zibin School of Computer Science and Engineering Sun Yat-Sen University Guangzhou China School of Software Engineering Sun Yat-Sen University Guangzhou China School of Software Engineering South China Normal University Foshan China

At present, the concept of metaverse has sparked widespread attention from the public to major industries. With the rapid development of blockchain and Web3 technologies, the decentralized metaverse ecology has attracted a large influx of users and capital. Due to the lack of industry standards and regulatory rules, the Web3-empowered metaverse ecosystem has witnessed a variety of financial crimes, such as scams, code exploit, wash trading, money laundering, and illegal services and shops. To this end, it is especially urgent and critical to summarize and classify the financial security threats on the Web3-empowered metaverse in order to maintain the long-term healthy development of its ecology. In this paper, we first outline the background, foundation, and applications of the Web3 metaverse. Then, we provide a comprehensive overview and taxonomy of the security risks and financial crimes that have emerged since the development of the decentralized metaverse. For each financial crime, we focus on three issues: a) existing definitions, b) relevant cases and analysis, and c) existing academic research on this type of crime. Next, from the perspective of academic research and government policy, we summarize the current anti-crime measurements and technologies in the metaverse. Finally, we discuss the opportunities and challenges in behavioral mining and the potential regulation of financial activities in the metaverse. The overview of this paper is expected to help readers better understand the potential security threats in this emerging ecology, and to provide insights and references for financial crime fighting. Copyright © 2022, The Authors. All rights reserved.

关键词： Taxonomies

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：