检索结果-内蒙古大学图书馆

Incorporating Option and Out-of-domain Knowledge for Multi-choice Machine Reading Comprehension 7

学校读者我要写书评

暂无评论

Incorporating Option and Out-of-domain Knowledge for Multi-c...

7th IEEE International Conference on Cloud computing and Intelligence Systems, CCIS 2021

作者： Xu, Yuan Shi, Shumin Huang, Heyan Beijing Institute of Technology School of Computer Science and Technology Beijing China Beijing Engineering Research Center of High Volume Language Information Processing and Cloud Computing Applications Beijing China

ISBN: (纸本)9781665441490

Multi-choice Machine Reading Comprehension (MRC) requires the model to select the correct answer from a set of answer candidates given the corresponding passage and question. Previous studies mainly focus on complex matching networks to model the relationship among options, passage and question. However, these models obtain little improvement over the powerful Pre-trained Language Models (PLMs). In this paper, we propose a simple method to incorporate option knowledge from PLMs and introduce out-of-domain knowledge by multi-task learning skillfully. Our approach obtains state-of-the-art results on Chinese multi-choice MRC dataset ReCO and also effectively improves the performance on C3. © 2021 IEEE.

关键词： Domain Knowledge

Fake Twitter Followers Detection using Machine Learning Approach

学校读者我要写书评

暂无评论

Fake Twitter Followers Detection using Machine Learning Appr...

Business Analytics for technology and Security (ICBATS), International Conference on

作者： Muhammad Zeshan Shabbir Iftikhar Naseer Shamim Akhter Muhammad Abubakar Ghassan F. Issa Muhammad Hassaan Mehmood Faculty of Information and Technology Comsats University Islamabad Pakistan Faculty of Computer Science & Information Technology The Superior University Lahore Pakistan School of Library & Information Sciences Minhaj University Lahore Pakistan Fast School of Computing National University of Computer and Emerging Sciences Lahore Pakistan School of Information Technology Skyline University College University City Sharjah Sharjah UAE School of Computer Science NCBA&E Lahore Pakistan Applied Science Research Center Applied Science Private Universit Amman Jordan

Fake user accounts on social media are a serious threat. Artificial intelligence can play an important role in getting rid of fake user accounts. Twitter has a problem with fake accounts that are run by automated bots. These bots are used for bad things like spamming, sabotaging trends, and getting more followers. In this research, publicly available information about Twitter users, such as their activity patterns, profile information, and tweets, to figure out how real each one is. Twitter’s API and other ways of getting information will be used to get user data. The proposed model consists of a machine learning technique, and it detects whether the user accounts are real or fake. Various algorithms such as logistic regression, long short-term memory, K-mean, and random forest are applied to evaluate the proposed model. The experimental results show that the random forest algorithm has obtained the highest accuracy 0.7557, a precision of 0.7277, and F1 of 0.7943 among all other algorithms.

关键词： Machine learning algorithms Social networking (online) Unsolicited e-mail Blogs Forestry Chatbots Market research

Fairness of Large Music Models: From a Culturally Diverse Perspective

学校读者我要写书评

暂无评论

Fairness of Large Music Models: From a Culturally Diverse Pe...

IEEE International Conference on Data science in Cyberspace (DSC)

作者： Qinyuan Wang Bruce Gu He Zhang Yunfeng Li School of Music University of Sydney NSW Australia Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center Qilu University of Technology (Shandong Academy of Sciences) Jinan China Shandong Provincial Key Laboratory of Computer Networks Shandong Fundamental Research Center for Computer Science Jinan China CNPIEC KEXIN Ltd Beijing China

ISBN: (数字)9798350391367

ISBN: (纸本)9798350391374

This paper explores the fairness of large music models in generating culturally diverse musical compositions, emphasizing the need for inclusivity and equity in AI-generated music. By analyzing popular datasets such as the Million Song Dataset and the Lakh MIDI Dataset, we identify a significant underrepresentation of non-Western musical elements. To address this, we adapt existing models to incorporate non-Western scales, rhythms, and instruments. The adapted models demonstrate a substantial improvement in generating culturally diverse music. Using AI-based synthetic listeners, we evaluate the impact of these adaptations on listener perception and engagement. Results indicate that music from the adapted models scores higher in terms of enjoyment, novelty, cultural resonance, and overall engagement compared to the original models. Our findings highlight the importance of cultural diversity in enhancing the user experience and ensuring ethical AI practices. The study also discusses challenges, such as dataset limitations and model adaptation complexity, and suggests directions for future research to further promote fairness and inclusivity in music AI. This work contributes to the development of more culturally aware and representative music generation systems, aiming to enrich the global musical landscape and foster greater appreciation of diverse musical traditions.

关键词： Adaptation models Ethics Instruments Data science Rhythm Data models User experience Resonance Cultural differences Artificial intelligence

Distributed Pairwise Protection for Security-Aware Mission Chains in UAV Networks

学校读者我要写书评

暂无评论

Distributed Pairwise Protection for Security-Aware Mission C...

World Congress on computing and Communication Technologies (WCCCT)

作者： Yuhang Zhao Guangming Duan Danyang Zheng Xiaojun Cao Chengzong Peng School of Cybersecurity (Xin Gu Industrial College) Chengdu University of Information Technology China SUGON Industrial Control and Security Center China School of Computing and Artificial Intelligence Southwest Jiaotong University China Department of Computer Science Georgia State University USA

ISBN: (数字)9798331512620

ISBN: (纸本)9798331512637

The unmanned aerial vehicles (UAVs) communication network exhibits significant potential in natural disaster management, with applications in flood relief and wildfire control. In these scenarios, UAVs can dynamically form mission chains (MCs) to collaboratively execute tasks such as real-time monitoring and post-disaster search and rescue. To address security challenges in these dynamic environments, we implement MCs as security-aware service function chains (SFCs). However, traditional SFC techniques are often inefficient and resource-intensive when applied to MCs in UAV networks, due to the networks’ dynamic nature and resource constraints. In this paper, we introduce and mathematically formulate a novel problem, termed the security-aware SFC distributed pairwise protection (SSFC-DPP) problem in UAV networks, which optimizes SFC protection against failures while balancing security and resource demands, and prove its NP-hardness. To tackle SSFC-DPP, we propose an efficient heuristic approach, the distributed pairwise node protection (DPNP) algorithm, integrating a security-resource ratio (SRR) factor and pairwise backup selection (PBS) technique. Extensive simulations show that DPNP reduces overall backup costs by 8.05% and 51.39% compared to two benchmark algorithms, respectively.

关键词： Wildfires Service function chaining Heuristic algorithms Benchmark testing Autonomous aerial vehicles Real-time systems Security Reliability Protection Vehicle dynamics

RETRACTED ARTICLE: Obstacles Uncovering System for Slender Pathways Using Unmanned Aerial Vehicles with Automatic Image Localization Technique

学校读者我要写书评

暂无评论

International Journal of Computational Intelligence Systems 2023年第1期16卷 1-14页

作者： Selvarajan, Shitharth Manoharan, Hariprasath Khadidos, Alaa O. Shankar, Achyut Khadidos, Adil O. Onyema, Edeh Michael Department of Computer Science Kebri Dehar University Kebri Dehar Ethiopia School of Built Environment Engineering and Computing Leeds Beckett University Leeds UK Department of Electronics and Communication Engineering Panimalar Engineering College Chennai India Department of Information Systems Faculty of Computing and Information Technology King Abdulaziz University Jeddah Saudi Arabia Center of Research Excellence in Artificial Intelligence and Data Science King Abdulaziz University Jeddah Saudi Arabia Secure Cyber Systems Research Group (SCSRG) WMG University of Warwick Coventry UK School of Computer Science Engineering Lovely Professional University Phagwara India Department of Information Technology Faculty of Computing and Information Technology King Abdulaziz University Jeddah Saudi Arabia Mathematics and Computer Science Coal City University Enugu Nigeria

In this study, unidentified flying machines are built with real-time monitoring in mid-course settings for obstacle avoidance in mind. The majority of the currently available methods are implemented as comprehensive monitoring systems, with significant success in monitored applications like bridges, railways, etc. So, the predicted model is developed exclusively for specific monitoring settings, as opposed to the broad conditions that are used by the current approaches. Also, in the design model, the first steps are taken by limiting the procedure to specific heights, and the input thrust that is provided for take up operation is kept to a minimum. Due to the improved altitudes, the velocity and acceleration units have been cranked up on purpose, making it possible to sidestep intact objects. In addition, Advanced Image Mapping Localization (AIML) is used to carry out the implementation process, which identifies stable sites at the correct rotation angle. Besides, Cyphal protocol integration improves the security of the data-gathering process by transmitting information gathered from sensing devices. The suggested system is put to the test across five different case studies, where the designed Unmanned aerial vehicle can able to detect 25 obstacles in the narrow paths in considered routs but existing approach can able to identify only 14 obstacle in the same routes.

关键词：

A GAN-Based Data Poisoning Attack Against Federated Learning Systems and Its Countermeasure

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Sun, Wei Gao, Bo Xiong, Ke Wang, Yuwei The Engineering Research Center of Network Management Technology for High Speed Railway of Ministry of Education School of Computer Science and Technology The Collaborative Innovation Center of Railway Traffic Safety Beijing Jiaotong University Beijing100044 China The Institute of Computing Technology Chinese Academy of Sciences Beijing100190 China

As a distributed machine learning paradigm, federated learning (FL) is collaboratively carried out on privately owned datasets but without direct data access. Although the original intention is to allay data privacy concerns, "available but not visible" data in FL potentially brings new security threats, particularly poisoning attacks that target such "not visible" local data. Initial attempts have been made to conduct data poisoning attacks against FL systems, but cannot be fully successful due to their high chance of causing statistical anomalies. To unleash the potential for truly "invisible" attacks and build a more deterrent threat model, in this paper, a new data poisoning attack model named VagueGAN is proposed, which can generate seemingly legitimate but noisy poisoned data by untraditionally taking advantage of generative adversarial network (GAN) variants. Capable of manipulating the quality of poisoned data on demand, VagueGAN enables to trade-off attack effectiveness and stealthiness. Furthermore, a cost-effective countermeasure named Model Consistency-Based Defense (MCD) is proposed to identify GAN-poisoned data or models after finding out the consistency of GAN outputs. Extensive experiments on multiple datasets indicate that our attack method is generally much more stealthy as well as more effective in degrading FL performance with low complexity. Our defense method is also shown to be more competent in identifying GAN-poisoned data or models. The source codes are publicly available at https://***/SSssWEIssSS/VagueGAN-Data-Poisoning-Attack-and-Its-Countermeasure. Copyright © 2024, The Authors. All rights reserved.

关键词： Cost effectiveness

Environment-Aware Dynamic Graph Learning for Out-of-Distribution Generalization

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Yuan, Haonan Sun, Qingyun Fu, Xingcheng Zhang, Ziwei Ji, Cheng Peng, Hao Li, Jianxin Beijing Advanced Innovation Center for Big Data and Brain Computing Beihang University China School of Computer Science and Engineering Beihang University China Key Lab of Education Blockchain and Intelligent Technology Guangxi Normal University China Department of Computer Science and Technology Tsinghua University China

Dynamic graph neural networks (DGNNs) are increasingly pervasive in exploiting spatio-temporal patterns on dynamic graphs. However, existing works fail to generalize under distribution shifts, which are common in real-world scenarios. As the generation of dynamic graphs is heavily influenced by latent environments, investigating their impacts on the out-of-distribution (OOD) generalization is critical. However, it remains unexplored with the following two major challenges: (1) How to properly model and infer the complex environments on dynamic graphs with distribution shifts? (2) How to discover invariant patterns given inferred spatio-temporal environments? To solve these challenges, we propose a novel Environment-Aware dynamic Graph LEarning (EAGLE) framework for OOD generalization by modeling complex coupled environments and exploiting spatiotemporal invariant patterns. Specifically, we first design the environment-aware EA-DGNN to model environments by multi-channel environments disentangling. Then, we propose an environment instantiation mechanism for environment diversification with inferred distributions. Finally, we discriminate spatio-temporal invariant patterns for out-of-distribution prediction by the invariant pattern recognition mechanism and perform fine-grained causal interventions node-wisely with a mixture of instantiated environment samples. Experiments on real-world and synthetic dynamic graph datasets demonstrate the superiority of our method against state-of-the-art baselines under distribution shifts. To the best of our knowledge, we are the first to study OOD generalization on dynamic graphs from the environment learning perspective. Copyright © 2023, The Authors. All rights reserved.

关键词： Graph neural networks

CroPrompt: Cross-task Interactive Prompting for Zero-shot Spoken Language Understanding

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Qin, Libo Wei, Fuxuan Chen, Qiguang Zhou, Jingxuan Huang, Shijue Si, Jiasheng Lu, Wenpeng Che, Wanxiang School of Computer Science and Engineering Central South University China Research Center for Social Computing and Information Retrieval Harbin Institute of Technology China Harbin Institute of Technology Shenzhen China Key Laboratory of Computing Power Network and Information Security Ministry of Education Qilu University of Technology Shandong Academy of Sciences China

Slot filling and intent detection are two highly correlated tasks in spoken language understanding (SLU). Recent SLU research attempts to explore zero-shot prompting techniques in large language models to alleviate the data scarcity problem. Nevertheless, the existing prompting work ignores the cross-task interaction information for SLU, which leads to sub-optimal performance. To solve this problem, we present the pioneering work of Cross-task Interactive Prompting (CroPrompt) for SLU, which enables the model to interactively leverage the information exchange across the correlated tasks in SLU. Additionally, we further introduce a multi-task self-consistency mechanism to mitigate the error propagation caused by the intent information injection. We conduct extensive experiments on the standard SLU benchmark and the results reveal that CroPrompt consistently outperforms the existing prompting approaches. In addition, the multi-task self-consistency mechanism can effectively ease the error propagation issue, thereby enhancing the performance. We hope this work can inspire more research on cross-task prompting for SLU. © 2024, CC BY-NC-SA.

关键词： Speech recognition

Text-Driven Video Style Transfer with State-Space Models: Extending StyleMamba for Temporal Coherence

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Li, Chao Park, Minsu Rossi, Cristina Li, Zhuang Dept. of Computer Science Eastern Asia Institute of Technology Beijing China AI Research Center Seoul National Technical Labs Seoul Korea Republic of School of Computing University of South Europe Rome Italy Department of Data Science Western Frontier University CA United States

StyleMamba has recently demonstrated efficient text-driven image style transfer by leveraging state-space models (SSMs) and masked directional losses. In this paper, we extend the StyleMamba framework to handle video sequences. We propose new temporal modules, including a Video State-Space Fusion Module to model inter-frame dependencies and a novel Temporal Masked Directional Loss that ensures style consistency while addressing scene changes and partial occlusions. Additionally, we introduce a Temporal Second-Order Loss to suppress abrupt style variations across consecutive frames. Our experiments on DAVIS and UCF101 show that the proposed approach outperforms competing methods in terms of style consistency, smoothness, and computational efficiency. We believe our new framework paves the way for real-time text-driven video stylization with state-of-the-art perceptual results. Copyright © 2025, The Authors. All rights reserved.

关键词： Video recording