Yi-Chun Hung
- yi-chunhung2028@u.northwestern.edu
- Mudd 3534, Northwestern University
Mr. Yi-Chun Hung is currently a PhD student at Northwestern CS. He received his M.S. degree in electrical and computer engineering from University of California, Los Angeles. He has conducted multiple projects related to computational imaging and computer vision. The outcomes of his research works were published into several international journals and proceeding conference papers. He has several patents in Taiwan and the US. He has also received several international awards and scholarships including J. Yang scholarship (2020), National Science Council Research Scholarship, Taiwan (2018) and IEEE Arctic Challenge Sponsorship (2018, only 6 winners over the world). Additionally, he is an action-oriented educator to inspire his juniors to realize creative ideas. To this end, he launched a semester-long cultivating program in 2017 to encourage juniors for original and innovative projects. The program benefited over 50 juniors and has been officially supported by the department. The program has been run by juniors for over 3 years.
Researches
Bio-inspired Computer Vision
All-Optical Diffractive Deep Neural Network
- Implemented and trained a broadband diffractive neural network that delivers superior performance on output quality and power efficiency compared to some of our previously demonstrated results.
- Proposed new loss function to accurately control the output spectral profile of a broadband diffractive network.
- Proposed an innovative modular design for mechanically assembling parts of diffractive networks with high precision and reconfigurability.
Terahertz Deep Learning Computed Tomography
- Designed the first supervised deep learning model based on terahertz time domain spectroscopy (THz-TDS) system to achieve the image resolution closed to diffraction limit.
- Established the precise THz-TDS dataset, containing over 600 gigabytes raw signal data.
- Utilized distributed computing system with automated hyperparameter tunning tool to improve the training efficiency.
Terahertz Compressed Sensing Imaging
- Built the terahertz compressed sensing optical system by a 532nm laser on a spatial light modulator.
- Implemented optimization based reconstruction algorithms, such as Hadamard matrix with Lasso.
- Collaborating with the professor in polymer material design to improve the modulation efficiency.
Terahertz Hyperspectral Imaging
- Formulated a terahertz data model for blind spectral unmixing algorithms.
- Designed the precise experiments to validate the assumptions in the formulated model.
- Collaborating with the professor in the optimization field to design the imaging reconstruction algorithm.
Seleted Interdisciplinary Projects
Material Parameters Extraction Among Terahertz Region
- Implementing the optimization based algorithm to extract the refractive index and absortion coefficient of the materials based on the Terahertz time domain system.
Learning-Based Autopiloted Cart
- Implemented lane following carts with object detection to avoid obstacles on Udoo Neo and K64F.
Interactive RPG Mobile Game iOS Application
- Implemented an IOS application for over 100 freshman welcome camp, including utilizing google map API, restful API, and data synchronizing with the self-build server.
Work Experience
Research Assistant, Yang Research Group
- Leading interdisciplinary projects relative to optics, numerical modeling, and optimization.
- Collaborating with professors from different fields, such as digital IC design, optimization, and polymer material.
Teaching Assistant, Embedded System Laboratory
- Organized the teaching material for image recongnition based on deep learning onto embedded platform.
- Implemented the lane following control system for the embedded cart based on image processing.
Research Assistant, Intelligent Hyperspectral Computing
- Proposed a hyhrid quantum-classical network for image restoration.
- Developed a model-based hyperspectral video motion deblur method.