A5100337347- Human Pose and Action Recognition
- Innovative Human-Technology Interaction
- Technology Use by Older Adults
- Advanced Graph Neural Networks
- Child Development and Digital Technology
- Data Management and Algorithms
- Domain Adaptation and Few-Shot Learning
- Adversarial Robustness in Machine Learning
- Blockchain Technology Applications and Security
- Advanced Vision and Imaging
- Multimodal Machine Learning Applications
- Face and Expression Recognition
- Explainable Artificial Intelligence (XAI)
- Advanced Database Systems and Queries
- Topic Modeling
- Human Motion and Animation
- Digital Mental Health Interventions
- Machine Learning and ELM
- Natural Language Processing Techniques
- Advanced Image and Video Retrieval Techniques
- Image and Signal Denoising Methods
- Neural Networks and Applications
- Model-Driven Software Engineering Techniques
- Handwritten Text Recognition Techniques
- Virtual Reality Applications and Impacts
Aalborg University
2024
Beijing Institute of Technology
2022-2024
McMaster University
2024
University of Minnesota
2019-2024
Simon Fraser University
2023-2024
Nanjing University of Finance and Economics
2022-2023
Shenyang University of Technology
2023
Shantou University
2022
Beijing University of Agriculture
2022
Megvii (China)
2022
Denoising diffusion models hold great promise for generating diverse and realistic human motions. However, existing motion largely disregard the laws of physics in process often generate physically-implausible motions with pronounced artifacts such as floating, foot sliding, ground penetration. This seriously impacts quality generated limits their real-world application. To address this issue, we present a novel physics-guided model (PhysDiff), which incorporates physical constraints into...
This article is focused on the adaptive event-based interval type-2 (IT-2) fuzzy security control problem for networked systems with multiple network attacks. In order to save resources, this proposes an improved event-triggered model, which can adjust threshold dynamically according change of current signal and previous triggering signal. addition, due external environment, impact randomly occurring output bias considered in systems. Taking effects attacks scheme into consideration, a class...
The pervasive availability of voice assistants may support children in finding answers to informational queries by removing the literacy requirements text search (e.g., typing, spelling). However, most such systems are not designed for specific needs and preferences struggle with understanding intent their questions. In our investigation, we observed 87 27 adults interacting three Wizard-of-Oz speech interfaces arrive at questions that required reformulation. We found many some help reach an...
Including older adults as full stakeholders in digital society.
Artificial intelligence algorithms have been used to enhance a wide variety of products and services, including assisting human decision making in high-stake contexts. However, these are complex trade-offs, notably between prediction accuracy fairness population subgroups. This makes it hard for designers understand design or services way that respects users' goals, values, needs. We proposed method help users explore algorithms, visualize their select with trade-offs consistent goals...
Examining the values inherent in papers published at IDC provides a lens to our research community and informs path of future needs opportunities. We conducted content analysis expressed all full between 2011 2019 survey with first authors 20% these papers. examine types contributions, qualities behaviors seeks support children, role child other stakeholders design process, theories that inform research, criteria guide technical choices. discuss contributions core value trends over past two...
Prior research has highlighted opportunities for technology to better support the tabletop game experience in offline and online settings, but little work focused on social aspect of gaming. We investigated collaborative aspects gaming unique context "social distancing" during 2020 COVID-19 pandemic shed light remote With a multi-method qualitative approach (including digital ethnography in-depth interviews), we empirically studied how people appropriate existing technologies adapt their...
We propose embodied scene-aware human pose estimation where we estimate 3D poses based on a simulated agent's proprioception and scene awareness, along with external third-person observations. Unlike prior methods that often resort to multistage optimization, non-causal inference, complex contact modeling interactions, our method is one-stage, causal, recovers global in environment. Since 2D observations are coupled the camera pose, disentangle use multi-step projection gradient defined...
Blockchain has received great attention in academia and industry due to its decentralization immutability. From the perspective of transaction processing, blockchain is a distributed shared ledger database with characteristics decentralization, traceability, transparency. These features ensure security blockchain’s reliability. However, because network requires complex consensus verification between users, it causes problems such as high cost data exchange low system throughput. Such are...
Siblings play a crucial and long-lasting role in family connections relationships. However, with the older sibling transitioning out of their parental home, maintaining close relationship can be challenging, especially if siblings have large age difference. We conducted diary interview study nine families China which spaced siblings, to identify design opportunities for technology better support communication connection needs. contribute HCI community three aspects. First, we an empirical...
Given the growing need for automatic 3D content creation pipelines, various representations have been studied to generate objects from a single image. Due its superior rendering efficiency, Gaussian splatting-based models recently excelled in both reconstruction and generation. splatting approaches image generation are often optimization-based, requiring many computationally expensive score-distillation steps. To overcome these challenges, we introduce an Amortized Generative framework (AGG)...
Virtual reality (VR) offers promise in education given its immersive and socially engaging nature, but it can pose challenges for educators when creating VR-specific content. VR videos function as a new educational tool content creation due to their affordability user-friendliness. However, little empirical research exists on how utilize associated pedagogy real classes. Our employed contextual inquiry, through in-person interviews online surveys with 11 instructors gain actionable insights...
The availability of massive vehicle trajectory data enables the modeling road-network constrained movement as travel-cost distributions rather than just single-valued costs, thereby capturing inherent uncertainty and enabling improved routing quality. Thus, stochastic has been studied extensively in edge-centric model, where such costs are assigned to edges a graph representation road network. However, this model still disregards important information trajectories fails capture dependencies...
The growing development and commercialization of Virtual Reality (VR) allow more children to get access this technology. VR features a new, emotional relevant experience with sense presence high interactivity. In paper, we present VWorld, an immersive system designed boost children's creativity computational thinking skill. VWorld enables create their own virtual world by putting 3D objects on miniature map, then explore the control chosen constructing program sequences. We design...
To better ground technical (systems) investigation and interaction design of cross-device experiences, we contribute an in-depth survey existing multi-device practices, including fragmented workflows across devices the way people physically organize configure their workspaces to support such activity. Further, this documents a historically significant moment transition new future remote work, trend dramatically accelerated by abrupt switch work-from-home (and having contend with demands...
In recent years, with the development of deep learning technology, computer vision and natural language processing have made significant progress, establishing relationship between has attracted more attention. The spatio-temporal images taken by satellites or aircrafts scene people other things are main focus area. Existing methods yielded excellent results in image–text matching, but there is still room for improvement effectively using coarse fine-grained information. We propose a method...
In order to further reduce the fuel consumption and emissions of parallel hybrid electric vehicle, first all, multi-objective optimization problem is converted into single-objective problem. Then logic threshold control parameters are optimized with particle swarm simulated annealing algorithm(PSOSA). The strategy separately used for three different test drive cycles (UDDC, EUDC JA1015) finally compared which not optimized. results show that FC, HC, CO NO <sub...
Community intergenerational mentorship offers an opportunity to address older adults' social isolation while providing valuable one-on-one or small group learning experiences for elementary school students. Current organizations that support this kind of engagement focus on in-person visits place the burden logistics and transportation adult. However, as adults become less independent aging, coming schools in person becomes more challenging. We present a qualitative analysis current...
While self-supervised representation learning (SSL) has proved to be effective in the large model, there is still a huge gap between SSL and supervised method lightweight model when following same solution. We delve into this problem find that prone collapse semantic space simply performing instance-wise contrast. To address issue, we propose relation-wise contrastive paradigm with Relation Knowledge Distillation (ReKD). introduce heterogeneous teacher explicitly mine information...
We present FIMO, an innovative dataset comprising formal mathematical problem statements sourced from the International Mathematical Olympiad (IMO) Shortlisted Problems. Designed to facilitate advanced automated theorem proving at IMO level, FIMO is currently tailored for Lean language. It comprises 149 statements, accompanied by both informal descriptions and their corresponding LaTeX-based proofs. Through initial experiments involving GPT-4, our findings underscore existing limitations in...
Prior work has explored various technology designs to support intergenerational communications and connections through remote activities such as reading or play. However, few works have these technologies outside the family settings. In this work, we aimed understand how can social connectedness activities, by investigating use of a tablet-based AR system among older adult volunteers students for reading. We developed based on insights from previous research, deployed in field, observed over...