A Figure of My Current Research Interest
Embedded Files
A Few Selected Projects and Topics
I am interested in exploring foundational methods for learning features and representations, hoping to provide support for various applications in deep learning, computer vision, and multimedia.
e.g.,
Text-Guided Image Invariant Feature Learning [ https://arxiv.org/abs/2503.13805 ]
Robust representation in self-supervised searning. [ https://arxiv.org/abs/2309.03360 ]
Learning the Padding in Convolutional Neural Networks. [ https://arxiv.org/abs/2301.0460 ]
Perspective Transformation Layer. [ https://arxiv.org/abs/2201.05706 ]
Deep Morphological Neural Networks. [ https://arxiv.org/abs/1909.01532 ]
I have been exploring a range of multimedia methodologies and specialized deep learning designs for various application domains. My focus includes developing innovative approaches that address unique challenges and enhance the applicability of deep learning in these fields.
Deep Learning-based Image Watermarking
Watermarking refers to covertly embedding information (i.e., a watermark) into a cover medium. By designing deep learning methods such as specialized training schemes and novel layers, this image watermarking project aims at adaptivity and robustness. One example downstream application is to extract a watermark from camera-resampled marked images, and the end users can scan any cover image for more information.
Funded: NSF CRII 2104267
AI CodeLab: Enhancing AI Programming Education
This project aims to develop and pilot the AI CodeLab system, a cross-college educational initiative designed to offer hands-on AI programming experiences across diverse disciplines at the University of Nebraska Omaha. AI CodeLab focuses on creating an inclusive, interdisciplinary learning environment, where students from different fields can acquire and apply AI skills.
The AI CodeLab emphasizes practical coding exercises and interactive learning modules. The system's lightweight examples allow students to engage with core AI concepts without needing advanced computational resources, making AI education accessible to a broader audience. By integrating real-world AI experiences into various courses, AI CodeLab aims to prepare students for future demands while fostering interdisciplinary collaboration and research.
Funded by: UNO Weitz Innovation and Excellence
Hazards Precursor Prediction
This project consists of multiple teams and explores a multi-dimensional approach to transforming tribal community resilience against extreme wind hazards.
Our team leverages historical data to employ cross-modality self-supervised learning, aiming to form a comprehensive understanding of meteorological phenomena. This serves as a "long-term foundation" for weather patterns. In addition, real-time data becomes pivotal for the continuous learning phase, acting as the "working memory" of the model. This phase applies the insights gained from self-supervised learning on historical data to current, real-time observations.
Funded: NSF 2431053
Smart Aviation: Intelligent Analytics for Flight Safety and Airport Operations
In this project, we aim to develop deep learning models for analyzing airport operational voice data. The challenges include the use of specialized terminologies and non-grammatical sentences in air traffic communication, as well as label scarcity and the need for real-time inference. To address these, we will develop and employ airport domain-specific pre-training and fine-tuning, air traffic data augmentation, self-supervised learning, and model compression techniques.
Funded: Nebraska University (NU) Collaboration Initiative
Deep Learning-based Foreign Object Debris Detection
Foreign Object Debris is any substance alien to an airport system that can cause damage. We leverage multidisciplinary techniques by integrating machine learning, computer vision, small unmanned aerial technology, and traditional airport operations protocols to develop a system that helps overcome the high-cost, low-efficiency, and technical challenges of Foreign Object Debris detection that a great number of small-scale airports are facing.
Funded:
NASA Nebraska Space Grant
Nebraska University (NU) Collaboration Initiative
Google Sites
Report abuse