I am a master’s student in ECE at the University of Florida.
I am confident that I will be an incomparable researcher due to my varied and strong research experience, along with my eagerness.
I had a program by the government that teaches overall machine learning and deep learning for 6 months. Also, I did an internship for 4 months. During my internship, I did a project related to self-supervised learning, semantic image segmentation, and 6D object detection. When I studied 6D object detection, I learned that measuring the depth between the camera and objects is very important to detect objects, especially transparent objects.
I took Fundamentals of Machine Learning, Neural Networks and Deep Learning, Image Processing and Computer Vision and Pattern Recognition in the ECE department. In the course project, for classification problems, I used MLP and CNNs for supervised learning and an Autoencoder that is used information-theoretic learning (ITL) regularization for unsupervised learning. Moreover, I improved the performance of the small object detection in remote sensing images using both super-resolution and sliced inference. With the above experience, I tried unsupervised learning for saliency object detection using those above methods.
Currently, I have interests in how I can apply computer vision to the Fabrication process and integrated circuits (ICs) while I am leading Semiconductor Device Fabrication Laboratory course as a teaching assistant. I want to try Autoencoder and Super-resolution methods in deep learning to get better resolution and denoised SEM images for better detection. Also, I want to use saliency detection and autoencoder methods to improve the performance of detecting counterfeit ICs. I have got an idea from a project that I am doing now. The project is making video playtime to be short including only important frames using saliency detection and autoencoder. I am sure that I can detect important parts and anomalies in ICs by applying those methods.