A5085160493- Complex Network Analysis Techniques
- Advanced Graph Neural Networks
- Anomaly Detection Techniques and Applications
- Video Analysis and Summarization
- Human Pose and Action Recognition
- Advanced Manufacturing and Logistics Optimization
- Traffic Prediction and Management Techniques
- Traffic and Road Safety
- Urban and Freight Transport Logistics
- Manufacturing Process and Optimization
- Recommender Systems and Techniques
- Assembly Line Balancing Optimization
- Advanced Vision and Imaging
- Autonomous Vehicle Technology and Safety
- Video Coding and Compression Technologies
- Opinion Dynamics and Social Influence
- Innovation Diffusion and Forecasting
- Advanced Image Processing Techniques
- Age of Information Optimization
- Bioinformatics and Genomic Networks
- Air Quality Monitoring and Forecasting
- Evacuation and Crowd Dynamics
- Optimization and Packing Problems
Tsinghua University
2021-2024
Nanjing University
2024
Dynamic link prediction is essential for a wide range of domains, including social networks, bioinformatics, knowledge bases, and recommender systems. Existing works have demonstrated that structural information temporal are two the most important this problem. However, existing either focus on modeling them independently or dynamics single scale, neglecting complex correlations among them. This paper proposes to model inherent evolving different scales dynamic prediction. Following idea, we...
The phenomenal success of the newly-emerging social e-commerce has demonstrated that utilizing relations is becoming a promising approach to promote platforms. In this new scenario, one most important problems predict value community formed by closely connected users in networks due its tremendous business value. However, few works have addressed problem because 1) novel setting and 2) challenging nature structure complex effects on To bridge gap, we develop Multi-scale Structure-aware...
For walking pedestrians, when they are blocked by obstacles or other adjust their speeds and directions to avoid colliding with them, which is called collision avoidance behavior. This behavior the most complex part of pedestrians' processes its modeling simulation keys realistic crowd simulation, serves as foundation for various applications. However, existing methods either lack representation power accurately model do not it explicitly, leads a poor level realism simulation. To realize we...
Generally speaking, we can easily specify many causal relationships in the prediction tasks of ubiquitous computing, such as human activity prediction, mobility and health prediction. However, most existing methods these fields failed to take advantage this prior knowledge. They typically make predictions only based on correlations data, which hinders performance real-world scenarios, because a distribution shift between training data testing generally exists. To fill gap, proposed...
Recovering the fine-grained working process of couriers is becoming one essential problems for improving express delivery systems because knowing detailed how accomplish their daily work facilitates analyzing, understanding, and optimizing procedure. Although coarse-grained courier trajectories waybill time data can be collected, this problem still challenging due to noisy with spatio-temporal biases, lacking ground truth couriers’ behaviors, complex correlations between behaviors. Existing...
Temporal Action Detection (TAD) aims to identify the action boundaries and corresponding category within untrimmed videos. Inspired by success of DETR in object detection, several methods have adapted query-based framework TAD task. However, these approaches primarily followed predict actions at instance level (i.e., each its center point), leading sub-optimal boundary localization. To address this issue, we propose a new Dual-level framework, namely DualDETR, detect from both instance-level...
Inter-frame modeling is pivotal in generating intermediate frames for video frame interpolation (VFI). Current approaches predominantly rely on convolution or attention-based models, which often either lack sufficient receptive fields entail significant computational overheads. Recently, Selective State Space Models (S6) have emerged, tailored specifically long sequence modeling, offering both linear complexity and data-dependent capabilities. In this paper, we propose VFIMamba, a novel...
With the prevalence of social media, rewarding users to invite their friends use an application or adopt a service has become one most successful designs in recent years. As spreads, diffusion process driven by incentivized invitation forms. However, despite extensive attention from both industry and academia, this newly-emerging remains under-explored. To bridge gap between academic research industrial practice, we set out examine influence hubs, i.e., people with considerable number ties,...
The two-dimensional bin packing problem (2DBP) is a critical optimization in the furniture production and glass cutting industries, where objective to cut smaller-sized items from minimum number of large standard-sized raw materials. In practice, factories manufacture hundreds customer orders (sets items) every day, relieve pressure management, common practice group into batches for production, ensuring that one order are same batch instead scattered across line. this work, we formulate as...