Dong-Hwan Jang

I am Dong-Hwan Jang, a first-year Ph.D. student in Computer Science at the University of Illinois Urbana-Champaign (UIUC). My research goal is to build robust and adaptive visual systems. This pursuit currently guides my work in robust and efficient 3D and 4D vision, where I focus on developing scalable scene representations that can generalize under distribution shifts. This research builds upon my previous projects, which involved designing adaptive network modules (such as the DynOPool pooling layer and an implicit deblurring module) and improving model robustness through weight merging.

Before starting my Ph.D., I worked as a research engineer at the AI Research Center, Samsung Advanced Institute of Technology (SAIT). I earned my M.S. in Electrical and Computer Engineering from Seoul National University, where I was advised by Professor Bohyung Han, and also interned at NAVER AI Lab with Dongyoon Han and Sangdoo Yun.

Email  /  CV  /  Google Scholar  /  Twitter  /  Linkedin

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.

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.

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.

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.

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.

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.

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.