A5101605408- Advanced Image Processing Techniques
- Advanced Vision and Imaging
- Topic Modeling
- Anomaly Detection Techniques and Applications
- Natural Language Processing Techniques
- Multimodal Machine Learning Applications
- Generative Adversarial Networks and Image Synthesis
- Neural and Behavioral Psychology Studies
- Image Processing Techniques and Applications
- Human Motion and Animation
- Data Visualization and Analytics
- Traffic Prediction and Management Techniques
- Network Security and Intrusion Detection
- Image and Signal Denoising Methods
- Traffic and Road Safety
- Advanced Text Analysis Techniques
- Image Enhancement Techniques
- Transcranial Magnetic Stimulation Studies
- Hate Speech and Cyberbullying Detection
- Data Management and Algorithms
- Human Mobility and Location-Based Analysis
- Sentiment Analysis and Opinion Mining
- Image and Video Quality Assessment
- Human Pose and Action Recognition
- Reinforcement Learning in Robotics
China Telecom (China)
2022-2025
Xiamen University
2024
Harbin Engineering University
2024
Dalian Maritime University
2022-2024
Shenzhen Polytechnic
2024
Kunming Medical University
2023
China Telecom
2022-2023
Wuhan University of Technology
2018-2023
University of Maryland, College Park
2022
Carnegie Mellon University
2021-2022
In the animation industry, cartoon videos are usually produced at low frame rate since hand drawing of such frames is costly and time-consuming. Therefore, it desirable to develop computational models that can automatically interpolate in-between frames. However, existing video interpolation methods fail produce satisfying results on data. Compared natural videos, possess two unique characteristics make difficult: 1) cartoons comprise lines smooth color pieces. The areas lack textures...
Mounting evidence suggests that response inhibition involves both proactive and reactive inhibitory control, yet its underlying neural mechanisms remain elusive. In particular, the roles of right inferior frontal gyrus (IFG) parietal lobe (IPL) in control are still under debate. This study aimed at examining causal role IFG IPL using transcranial direct current stimulation (tDCS) stop signal task. Twenty-two participants completed three sessions task, anodal tDCS IFG, IPL, or primary visual...
Siyao Li, Deren Lei, Pengda Qin, William Yang Wang. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and 9th International Joint (EMNLP-IJCNLP). 2019.
Grounded Situation Recognition (GSR) aims to generate structured semantic summaries of images for "human-like" event understanding. Specifically, GSR task not only detects the salient activity verb (e.g. buying), but also predicts all corresponding roles agent and goods). Inspired by object detection image captioning tasks, existing methods typically employ a two-stage framework: 1) detect verb, then 2) predict based on detected verb. Obviously, this illogical framework constitutes huge...
Over the past two decades, sequential recommendation systems have garnered significant research interest, driven by their potential applications in personalized product recommendations. In this article, we seek to explicitly model an algorithm based on Internet of Things (IoT) data predict next cell reached user equipment (UE). This exploits UE embedding and combining visit time interval information, uses sliding window sampling process more trajectory data. Furthermore, use attention...
In the past few years, there has been a surge of interest in multi-modal problems, from image captioning to visual question answering and beyond. this paper, we focus on hate speech detection memes wherein pose an interesting fusion problem. We aim solve Facebook Meme Challenge \cite{kiela2020hateful} which aims binary classification problem predicting whether meme is hateful or not. A crucial characteristic challenge that it includes "benign confounders" counter possibility models...
Yi Cheng, Siyao Li, Bang Liu, Ruihui Zhao, Sujian Chenghua Lin, Yefeng Zheng. Proceedings of the 59th Annual Meeting Association for Computational Linguistics and 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.
Current studies on traffic crash prediction mainly focus the frequency and severity of freeways or arterials. However, collision type for urban expressway is rarely considered. Meanwhile, with rapid development systems in China recent years, safety problems have attracted more attention. In addition, characteristics are considered to be a potentially important predictor accidents; however, their impact crashes has been controversial. Therefore, predicting model considering types proposed...
This paper reviews the NTIRE 2020 challenge on video quality mapping (VQM), which addresses issues of from source domain to target domain. The includes both a supervised track (track 1) and weakly-supervised 2) for two benchmark datasets. In particular, 1 offers new Internet benchmark, requiring algorithms learn map more compressed videos less in training manner. 2, are required one device another when their varies substantially weakly- aligned pairs available. For 1, total 7 teams competed...
Cross-border transactions have been more and popular around the world. However, current cross-border still risks challenges, e.g., differences in regulation policies unbalanced profits of banks. To address this critical issue, we construct a new framework for transaction system with support blockchain technology. In paper, propose consortium system, namely asymmetric (ACB), to ensure implementation transactions. Different from traditional blockchain, could supernode regulate all timely....
Abstract Fundamental theories of human cognition have long posited that the short‐term maintenance actions is supported by one “core knowledge” systems visual cognition, yet its neural substrates are still not well understood. In particular, it unclear whether memory (VSTM) has distinct or, as proposed spatio‐object architecture VSTM, shares them with VSTM objects and spatial locations. two experiments, we tested these competing hypotheses directly contrasting for those Our results showed...
Videos contain various types and strengths of motions that may look unnaturally discontinuous in time when the recorded frame rate is low. This paper reviews first AIM challenge on video temporal super-resolution (frame interpolation) with a focus proposed solutions results. From low-frame-rate (15 fps) sequences, participants are asked to submit higher-frame-rate (60 sequences by estimating temporally intermediate frames. We employ REDS_VTSR dataset derived from diverse videos captured...
An important task in the study of Natural Language Processing (NLP) is analysis movie reviews. It finishes classifying review texts into sentiment, such as positive, negative or neutral sentiment. Previous works mainly follow pipeline LSTM (Long Short-Term Memory Network). The network model a variant Recurrent Neural Network (RNN) and particularly suitable for processing natural language texts. Though existing LSTM-based have improved performance significantly, we argue that most them deal...
We introduce a novel task within the field of 3D dance generation, termed accompaniment, which necessitates generation responsive movements from partner, "follower", synchronized with lead dancer's and underlying musical rhythm. Unlike existing solo or group tasks, duet scenario entails heightened degree interaction between two participants, requiring delicate coordination in both pose position. To support this task, we first build large-scale diverse interactive dataset, DD100, by recording...
Video interpolation is an important problem in image manipulation, which has drawn increased interests from the vision and graphics communities. In this work, we apply quadratic video algorithm to VTSR challenge of Advances Image Manipulation (AIM) 2019, introduce a joint finetuning scheme exploit more training data. We provide concise description model present detailed analysis results challenge. Extensive experiments demonstrate that our network generates high-quality outperforms...
The promise of visualization recommendation systems is that analysts will be automatically provided with relevant and high-quality visualizations reduce the work manual exploration or chart creation. However, little research to date has focused on what value in design recommendations. We interviewed 18 public health sector explored how they made sense a popular in-domain dataset1 service generating recommend others. also interacted corpus both automatically- manually-generated...
Malicious software or misbehaving applications have the potential to trigger signaling storms on mobile networks, leading battery drainage devices and causing bandwidth overuse at cell level. Additionally, these may result in an excessive load within operator's infrastructure. This paper uses a combination of time series prediction, adaptive threshold, anomaly detection algorithms predict storms. Whether storm will be triggered future can determined based fluctuation pattern data. Our method...
Existing correspondence datasets for two-dimensional (2D) cartoon suffer from simple frame composition and monotonic movements, making them insufficient to simulate real animations. In this work, we present a new 2D animation visual dataset, AnimeRun, by converting open source three-dimensional (3D) movies full scenes in style, including simultaneous moving background interactions of multiple subjects. Our analyses show that the proposed dataset not only resembles anime more image...