I'm

Vincent-Daniel Yun

Neural Network Researcher

M.S student at University of Southern California





Announcements


[05.2025] I will be joining the University of Southern California for the MSCS program starting in Fall 2025
[05.2025] Two papers were submitted to a ML conference (Core Rank: A)
Sharpness-Aware Minimization with Z-Score Gradient Filtering for Neural Networks [Paper]
Spectral and Temporal Denoising for Differentially Private Optimization [Paper]
[05.2025] One paper was accepted at Neural Computing and Application (IF: 4.5)
Stochastic Gradient Sampling for Enhancing Neural Networks Training [Paper]
[04.2025] One paper was submitted to a ML conference (Core Rank: B)
Revisiting 16-bit Neural Network Training: A Practical Approach for Resource-Limited Learning [Paper]
[01.2025] Launched the Open Neural Network Research Lab, supported by the Brian Impact Foundation
Eleven researchers from global research institutions joined the lab [Read More]
[12.2024] One paper was accepted at AAAI 2025 Workshops
ZNorm: Z-Score Gradient Normalization Accelerating Skip-Connected Network Training without Architectural Modification [Paper]
[11.2024] One paper was accepted at IEEE BigData 2024
Mitigating Gradient Overlap in Deep Residual Networks with Gradient Normalization for Improved Non-Convex Optimization [Paper]
[07.2024] One paper was accepted at European Space Agency (ESA) SPAICE 2024
Analysis and Predictive Modeling of Solar Coronal Holes Using Computer Vision and ARIMA-LSTM Networks [Paper]
[04.2024] One paper was accepted at CVPR 2024 Workshops
Uncertainty Estimation for Tumor Prediction with Unlabeled Data [Paper]
[03.2024] One paper was accepted at IJCNN 2024
Robust Neural Pruning with Gradient Sampling Optimization for Residual Neural Networks [Paper]


See More









About Me

Hi, I’m Vincent-Daniel Yun, also known as Daniel. I’m a Computer Science master’s student at the University of Southern California. Before USC, I studied Computer Science and Applied Mathematics at Stony Brook University in New York, where I worked as a research assistant under Professor Chao Chen at the Biomedical Informatics Center (BMI), focusing on uncertainty estimation in neural networks for breast cancer tumor prediction in digital pathology. Prior to BMI, I conducted research on neural network quantization under Professors Zhoulai Fu and Francois Rameau.

I’m passionate about neural network theory, learning theory, and optimization, particularly enhancing gradient descent and improving neural network generalization. My interests also include quantization, pruning, and neural architecture search (NAS) to accelerate training and inference while preserving performance, with a broad focus on advancing neural network architectures and theoretical frameworks through research.

  • Neural Networks
  • Learning Theory
  • Optimization
  • NAS
  • Theoretical Deep Learning
  • Uncertainty Estimation
Downlaod CV




Research Experiences

BMI Center at Stony Brook University

09.2023 - 05.2025
  • Research Assistant at Biomedical Informatics Center
  • Advisor: Prof. Chao Chen
  • Conducted research to enhance uncertainty estimation in breast cancer tumor prediction for digital pathology. Applied Consistency Ranking Loss to neural network training, leveraging unlabeled patch data to improve uncertainty estimation. Provided the trained model to Mayo Clinic to support ongoing research.

DNI Lab at Stony Brook University

05.2022 - 08.2023
  • Research Assistant at Data and Intelligent Computing Lab
  • Co-Advisor: Prof. Zhoulai Fu and Prof. Francois Rameau
  • Implemented a standalone 16-bit neural networks, optimizing computational efficiency, reducing training time by up to 2.2× while maintaining accuracy comparable to 32-bit models.

Space Team at Republic of Korea Air Force

08.2020 - 05.2022
  • Deep Learning Engineer at Space Team
  • Supervisor: Captain Jungmin Shin
  • Developed AI driven predictive models for solar coronal hole analysis, leveraging LSTM and ARIMA to forecast space weather patterns and their potential impact. Also developed a sunspot detection and tracking model, and presented the research conducted with the Defense Acquisition Program Administration of Korea at the European Space Agency in 2024.

Lead Lab at Stony Brook University

08.2019 - 08.2020
  • Research Assistant at Lead Lab
  • Advisor: Prof. Pravin Pawar
  • Developed an IoT and cloud-based energy prediction system, leveraging time-series forecasting models (ARIMA, VAR, and RNN) to optimize household electricity consumption.

Education

University of Southern California

08.2025 - 05.2027
  • Master of Science in Computer Science
  • Advisor:

Stony Brook University

01.2018 - 05.2025
  • Bachelor of Science in Applied Mathematics
  • Bachelor of Science in Computer Science
  • Advisors: Prof. Chao Chen
  • Co-advisors: Prof. Zhoulai Fu and Prof. Francois Rameau

Research Areas

Neural Networks

I research efficient neural network architectures, focusing on quantization, pruning, and neural architecture search (NAS). My work optimizes models for faster training and inference while maintaining performance.

Optimization

My optimization research centers on enhancing gradient descent algorithms to improve neural network training. I explore learning theory to develop methods that accelerate convergence while preserving performance.

Uncertainty Estimation

I am interested in advancing uncertainty estimation in neural networks to improve model reliability. My research explores novel techniques for quantifying uncertainty, such as confidence calibration and ranking-based methods.