检索结果-内蒙古大学图书馆

Revisiting Checkpoint Averaging for Neural Machine Translation

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Gao, Yingbo Herold, Christian Yang, Zijian Ney, Hermann Human Language Technology and Pattern Recognition Group Computer Science Department RWTH Aachen University Germany

Checkpoint averaging is a simple and effective method to boost the performance of converged neural machine translation models. The calculation is cheap to perform and the fact that the translation improvement almost comes for free, makes it widely adopted in neural machine translation research. Despite the popularity, the method itself simply takes the mean of the model parameters from several checkpoints, the selection of which is mostly based on empirical recipes without many justifications. In this work, we revisit the concept of checkpoint averaging and consider several extensions. Specifically, we experiment with ideas such as using different checkpoint selection strategies, calculating weighted average instead of simple mean, making use of gradient information and fine-tuning the interpolation weights on development data. Our results confirm the necessity of applying checkpoint averaging for optimal performance, but also suggest that the landscape between the converged checkpoints is rather flat and not much further improvement compared to simple averaging is to be obtained. Copyright © 2022, The Authors. All rights reserved.

关键词： Neural machine translation

Let's Influence Algorithms Together: How Millions of Fans Build Collective Understanding of Algorithms and Organize Coordinated Algorithmic Actions

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Xiao, Qing Zheng, Yuhang Fan, Xianzhe Zhang, Bingbing Lu, Zhicong Human-Computer Interaction Institute Carnegie Mellon University PittsburghPA United States University of Amsterdam Amsterdam Netherlands Tsinghua University Beijing China School of Journalism and Mass Communication University of Iowa Iowa CityIA United States Department of Computer Science George Mason University FairfaxVA United States

Previous research pays attention to how users strategically understand and consciously interact with algorithms but mainly focuses on an individual level, making it difficult to explore how users within communities could develop a collective understanding of algorithms and organize collective algorithmic actions. Through a two-year ethnography of online fan activities, this study investigates 43 core fans who always organize large-scale fans collective actions and their corresponding general fan groups. This study aims to reveal how these core fans mobilize millions of general fans through collective algorithmic actions. These core fans reported the rhetorical strategies used to persuade general fans, the steps taken to build a collective understanding of algorithms, and the collaborative processes that adapt collective actions across platforms and cultures. Our findings highlight the key factors that enable computer-supported collective algorithmic actions and extend collective action research into the large-scale domain targeting algorithms. Copyright © 2024, The Authors. All rights reserved.

关键词： Consensus algorithm

What Makes Im Users (Un)Responsive: An Empirical Investigation for Understanding Im Responsiveness

学校读者我要写书评

暂无评论

SSRN

SSRN 2022年

作者： Lee, Hao-Ping Chiang, Yi-Shyuan Chou, Yu-Ling Lin, Kung-Pai Chang, Yung-Ju Department of Computer Science National Yang Ming Chiao Tung University Hsinchu Taiwan Human-Computer Interaction Institute Carnegie Mellon University PittsburghPA United States Department of Computer Science University of Illinois Urbana-Champaign UrbanaIL United States

Instant messaging (IM) communication has been widely studied due to its prevalence in our everyday communication. Numerous factors that contribute to (un)responsiveness have been identified. Yet an integrated view of the factors that influence IM responsiveness remains absent. This paper reports qualitative findings from interviews with 46 IM users, and identifies five main elements underlying IM users’ response decisions: response habits, need fulfillment, perceived obligation, perceived readiness/suitability, and pace/rhythm coordination. To the best of our knowledge, this is the first integrated view of the key elements underlying IM responsiveness. Among these elements, we particularly highlight that regarding the influence of contextual factors on IM responsiveness, what may matters more is IM users’ perceptions of their readiness and suitability of the contexts, rather than the contexts’ objective properties. We also uncovered pace/rhythm coordination as a crucial factor behind IM responsiveness, which has been little discussed in the literature. Responsiveness, as our findings show, was often not a consequence, but a manifestation, of users’ pursuit of specific responsiveness-related aims such as ongoing shaping and sustaining of a dyad’s pace and rhythm. © 2022, The Authors. All rights reserved.

关键词：

Learning Co-Speech Gesture for Multimodal Aphasia Type Detection

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Lee, Daeun Son, Sejung Jeon, Hyolim Kim, Seungbae Han, Jinyoung Department of Applied Artificial Intelligence Sungkyunkwan University Seoul Korea Republic of Department of Human-AI Interaction Sungkyunkwan University Seoul Korea Republic of Department of Computer Science & Engineering University of South Florida TampaFL United States

Aphasia, a language disorder resulting from brain damage, requires accurate identification of specific aphasia types, such as Broca’s and Wernicke’s aphasia, for effective treatment. However, little attention has been paid to developing methods to detect different types of aphasia. Recognizing the importance of analyzing co-speech gestures for distinguish aphasia types, we propose a multimodal graph neural network for aphasia type detection using speech and corresponding gesture patterns. By learning the correlation between the speech and gesture modalities for each aphasia type, our model can generate textual representations sensitive to gesture information, leading to accurate aphasia type detection. Extensive experiments demonstrate the superiority of our approach over existing methods, achieving state-of-the-art results (F1 84.2%). We also show that gesture features outperform acoustic features, highlighting the significance of gesture expression in detecting aphasia types. We provide the codes for reproducibility purposes1 Copyright © 2023, The Authors. All rights reserved.

关键词： Damage detection

User-Driven Value Alignment: Understanding Users’ Perceptions and Strategies for Addressing Biased and Discriminatory Statements in AI Companions

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Fan, Xianzhe Xiao, Qing Zhou, Xuhui Pei, Jiaxin Sap, Maarten Lu, Zhicong Shen, Hong Tsinghua University Beijing China Human-Computer Interaction Institute Carnegie Mellon University PittsburghPA United States Language Technologies Institute Carnegie Mellon University PittsburghPA United States Stanford University StanfordCA United States Department of Computer Science George Mason University FairfaxVA United States

Large language model-based AI companions are increasingly viewed by users as friends or romantic partners, leading to deep emotional bonds. However, they can generate biased, discriminatory, and harmful outputs. Recently, users are taking the initiative to address these harms and re-align AI companions. We introduce the concept of user-driven value alignment, where users actively identify, challenge, and attempt to correct AI outputs they perceive as harmful, aiming to guide the AI to better align with their values. We analyzed 77 social media posts about discriminatory AI statements and conducted semi-structured interviews with 20 experienced users. Our analysis revealed six common types of discriminatory statements perceived by users, how users make sense of those AI behaviors, and seven user-driven alignment strategies, such as gentle persuasion and anger expression. We discuss implications for supporting user-driven value alignment in future AI systems, where users and their communities have greater agency. Copyright © 2024, The Authors. All rights reserved.

关键词：

Bridging Borders, Breaking Biases: Envisioning Technologies to Support North Korean Defectors in South Korea 25

学校读者我要写书评

暂无评论

Bridging Borders, Breaking Biases: Envisioning Technologies ...

2025 CHI Conference on human Factors in Computing Systems, CHI 2025

作者： Noh, Hayoun Jo, Hyunah Wang, Ge Van Kleek, Max Kang, Younah Department of Computer Science University of Oxford Oxford United Kingdom Yonsei University Seoul South Korea Institute for Human-Centered Artificial Intelligence (HAI) Stanford University Stanford CA United States Information and Interaction Design Yonsei University Seoul South Korea

ISBN: (纸本)9798400713941

North Korean defectors (NKDs) face significant challenges when transitioning to South Korean society. Leaving their homes permanently and adapting to a new, digitally connected environment for the first time presents difficulties, compounded by the pervasive stigma associated with their identities. Although technology alone cannot solve these issues, it can play a role in easing their transition. In this study, we conducted eight speculative co-creation sessions with 22 NKDs to identify their main challenges and envision potential technological interventions. We propose conceptualization of thirteen technologies aimed at addressing key issues NKDs face related to identity stigma, disconnection from their past, and challenges of adapting to a highly digital society. Through this empirical research on underrepresented populations undergoing significant life transitions, we provide insights into how future technologies can support other marginalized individuals as they navigate pervasive stigma and establish new lives in a digital society. © 2025 Copyright held by the owner/author(s).

关键词： Co-Creation Workshops Envisioning Technologies Future Technologies Marginalized Individuals North Korean Defector

Support Autonomy: Exploring Player Perspectives on AI-Supported Onboarding in Video Games 25

学校读者我要写书评

暂无评论

Support Autonomy: Exploring Player Perspectives on AI-Suppor...

2025 CHI Conference on human Factors in Computing Systems, CHI 2025

作者： Choong, Lydia Cmentowski, Sebastian Kukshinov, Eugene Tu, Joseph Nacke, Lennart E. Hci Games Group Stratford School of Interaction Design and Business University of Waterloo Waterloo ON Canada Cheriton School of Computer Science University of Waterloo Canada Industrial Design Eindhoven University of Technology Eindhoven Netherlands Department of Systems Design Engineering University of Waterloo Canada

ISBN: (纸本)9798400713941

Video game onboarding faces the challenge of teaching game mechanics in a fun and engaging way. Artificial intelligence (AI) solutions have become a quick fix to help users understand technology. However, little is known about how AI supports player onboarding in video games. To address this knowledge gap, this research explores player perspectives on AI-supported onboarding. We conducted a qualitative user study (n = 20) to investigate player expectations, attitudes, and concerns about AI-supported learning experiences. Players learn primarily through the lived experience of a game and value personalized guidance during onboarding. Participants emphasized the importance of maintaining control over how AI is used during onboarding and the freedom to choose their support level. Our results suggest that players want future AI-supported onboarding systems to prioritize their agency, encourage active learning, and maintain transparency throughout the learning process. We contribute to game design research by proposing balanced, player-centric AI-supported onboarding experiences in video games. © 2025 Copyright held by the owner/author(s).

关键词： artificial intelligence onboarding player experience video games

A Passive Variable Impedance Control Strategy with Viscoelastic Parameters Estimation of Soft Tissues for Safe Ultrasonography

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Beber, Luca Lamon, Edoardo Nardi, Davide Fontanelli, Daniele Saveriano, Matteo Palopoli, Luigi Department of Industrial Engineering Università di Trento Trento Italy Department of Information Engineering and Computer Science Università di Trento Trento Italy Human-Robot Interfaces and Interaction Istituto Italiano di Tecnologia Genoa Italy

In the context of telehealth, robotic approaches have proven a valuable solution to in-person visits in remote areas, with decreased costs for patients and infection risks. In particular, in ultrasonography, robots have the potential to reproduce the skills required to acquire high-quality images while reducing the sonographer’s physical efforts. In this paper, we address the control of the interaction of the probe with the patient’s body, a critical aspect of ensuring safe and effective ultrasonography. We introduce a novel approach based on variable impedance control, allowing the real-time optimisation of compliant controller parameters during ultrasound procedures. This optimisation is formulated as a quadratic programming problem and incorporates physical constraints derived from viscoelastic parameter estimations. Safety and passivity constraints, including an energy tank, are also integrated to minimise potential risks during human-robot interaction. The proposed method’s efficacy is demonstrated through experiments on a patient’s dummy torso, highlighting its potential for achieving safe behaviour and accurate force control during ultrasound procedures, even in cases of contact loss. Copyright © 2023, The Authors. All rights reserved.

关键词： Parameter estimation

Congenital hydrocephalus:a review of recent advances in genetic etiology and molecular mechanisms

学校读者我要写书评

暂无评论

Military Medical Research 2024年第6期11卷 942-958页

作者： Xiu-Yun Liu Xin Song Marek Czosnyka Chiara Robba Zofia Czosnyka Jennifer Lee Summers Hui-Jie Yu Guo-Yi Gao Peter Smielewski Fang Guo Mei-Jun Pang Dong Ming Medical School Tianjin UniversityTianjin 300072China State Key Laboratory of Advanced Medical Materials and Devices Tianjin UniversityTianjin 300072China Haihe Laboratory of Brain-Computer Interaction and Human-Machine Integration Tianjin 300380China School of Pharmaceutical Science and Technology Tianjin University300072 TianjinChina Department of Clinical Neurosciences Addenbrooke’s HospitalUniversity of CambridgeCambridge CB20QQUK San Martino Policlinico Hospital IRCCS for Oncology and Neuroscience16132 GenoaItaly Department of Anesthesiology and Critical Care Medicine Johns Hopkins UniversityBaltimoreMD 21287USA Department of Neurosurgery Tianjin Medical University General HospitalTianjin 300052China Department of Neurosurgery Beijing Tiantan HospitalCapital Medical UniversityBeijing 100070China Department of Neurosurgery Tianjin Huanhu HospitalTianjin 300350China

The global prevalence rate for congenital hydrocephalus(CH)is approximately one out of every five hundred births with multifaceted predisposing factors at *** influences stand as a major contributor to CH pathogenesis,and epidemiological evidence suggests their involvement in up to 40%of all cases observed *** about an individual’s genetic susceptibility can significantly improve prognostic precision while aiding clinical decision-making ***,the precise genetic etiology has only been pinpointed in fewer than 5%of human *** occurrences of CH cases are required for comprehensive gene sequencing aimed at uncovering additional potential genetic loci.A deeper comprehension of its underlying genetics may offer invaluable insights into the molecular and cellular basis of this brain *** review provides a summary of pertinent genes identified through gene sequencing technologies in humans,in addition to the 4 genes currently associated with CH(2 X-linked genes L1CAM and AP1S2,2 autosomal recessive MPDZ and CCDC88C).Others predominantly participate in aqueduct abnormalities,ciliary movement,and nervous system *** prospective CH-related genes revealed through animal model gene-editing techniques are further outlined,focusing mainly on 4 pathways,namely cilia synthesis and movement,ion channels and transportation,Reissner’s fiber(RF)synthesis,cell apoptosis,and ***,the proper functioning of motile cilia provides significant impulsion for cerebrospinal fluid(CSF)circulation within the brain ventricles while mutations in cilia-related genes constitute a primary cause underlying this *** far,only a limited number of CH-associated genes have been identified in *** integration of genotype and phenotype for disease diagnosis represents a new trend in the medical *** models provide insights into the pathogenesis of CH and contribute to our understanding of its association w

关键词： Congenital hydrocephalus Genetic disease Central system Cilia Ventricle