Dong-Hwan Jang
I am Dong-Hwan Jang, a recent Master's graduate in Electrical and Computer Engineering from Seoul National University, where I was mentored by Professor Bohyung Han. After graduation, I worked with Dongyoon Han and Sangdoo Yun during my internship at NAVER AI Lab. Now, I am a research engineer at the AI Research Center, Samsung Advanced Institute of Technology (SAIT).
As a machine learning researcher, my goal is to minimize human inductive bias in models for robust performance across diverse environments. Initially, I focused on dynamic neural networks, optimizing network structures and operations. Currently, my research focuses on robust fine-tuning using linear mode connectivity to enhance model robustness.
Email  / 
CV  / 
Google Scholar  / 
Twitter  / 
Linkedin
|
|
Education
• Visiting Scholar, Carnegie Mellon University (Sep 2022 - Feb 2023) · Full-time, Pittsburgh, PA · AI project, fully funded by the Korean Government
• Master's Student, ECE (Computer Vision), Seoul National University (Sep 2020 - Aug 2023) · Research advised by Prof. Bohyung Han
• Bachelor's Degree, ECE / MOT, Seoul National University (Mar 2013 - Aug 2020) · 1st out of 35, Summa Cum Laude
|
|
Model Stock: All we need is just a few fine-tuned models,
Dong-Hwan Jang,
Sangdoo Yun,
Dongyoon Han
ECCV, 2024 (oral, Top 8.35% among accepted papers).
arxiv /
github /
article (marktechpost)
Thanks to our insights in the fine-tuned weight space, fine-tuning a few models (i.e., only two) can lead to superior merged weights (closer to the center of a weight space) without merging many fine-tuned models under extensive parameter searches like Model Soup.
|
|
Merge and Bound: Direct Manipulations on Weights for Class Incremental Learning
Taehoon Kim,
Dong-Hwan Jang,
Bohyung Han
CVPR Workshop on Continual Learning in Computer Vision, 2024
We introduce Merge-and-Bound (M&B), an innovative approach for Class Incremental Learning that optimizes model weights through two merging techniques: inter-task and intra-task weight merging, alongside a bounded update to prevent catastrophic forgetting. Without altering architectures or objectives, M&B integrates into existing methods, showing superior performance on CIL benchmarks against top competitors.
|
|
Pooling Revisited: Your Receptive Field is Suboptimal
Dong-Hwan Jang,
Sanghyeok Chu,
Joonhyuk Kim,
Bohyung Han
CVPR, 2022
We propose DynOPool, a learnable pooling layer that finds the optimal scale factors and receptive fields of intermediate feature maps. It adapts feature map sizes and shapes for enhanced accuracy and efficiency in tasks like image classification and semantic segmentation. This innovative approach significantly improves deep neural network optimization with minimal computational overhead.
|
|
DS4C Patient Policy Province Dataset: a Comprehensive COVID-19 Dataset for Causal and Epidemiological Analysis
Jimi Kim*,
Seojin Jang*,
Joong Kun Lee*,
Dong-Hwan Jang* (* indicates equal contributions)
NeurIPS Workshop on Causal Discovery & Causality-Inspired Machine Learning, 2020
project page
/
article (korean)
(auto-translated)
We present DS4C South Korea Patient, Policy, and Provincial data (DS4C-PPP dataset). The dataset contains comprehensive data that could be used for causal analysis, such as per-patient symptom onset and confirmed date, travel frequency, hospital accessibility, and 61 preventative policies enacted in South Korea.
|
|
DIFF: Deblurring Implicit Feature Function
We propose spatially-variant motion deblur network based on the implicit neural representation. A spatially-variant deblurring network takes deformed features and their offsets as inputs.
U.S. Patent Application Number: 17/973,809 (in progress)
|
|
Dynamic Spatially-Adaptive Modulation for image Dehazing
We propose a novel framework of the dynamic spatially-adaptive modulation for image dehazing.
The proposed algorithm introduces a selection module that conditionally determine the necessary modulation pathways in a bottom-up manner by providing a loss function optimizing both accuracy and efficiency.
|
|
DepthFinder: Universal Image Restoration based on Adaptive Inference
We adaptively find the appropriate number of the residual blocks according to the severity and distortion type of the input in universal image restoration task.
|
Scholarships & Award
• Korean Government Scholarship for Overseas Study (2023-2024)
 - Covers USD 40,000 support per year. Only 64 students are selected in all fields in Korea.
• Hyundai OnDream Global Scholarship Award (2022)
 - Award Prize - around USD 2,350
 - For the paper “Pooling Revisited: Your Receptive Field is Suboptimal” at CVPR 2022
• Hyundai OnDream Future Technology Scholarship (2021-2022)
 - Covers full tuition & financial support
|
Teaching Experiences
• Teaching Assistant for 430.329: Introduction to Algorithms at Seoul National University (Fall 2020)
• Teaching Assistant for Samsung AI Expert Course at Seoul National University (July 2019)
• Teaching Assistant for Hyundai Motors AI Expert Course at Seoul National University (Jan 2019)
|
|