Sihyeon Kim [김시현 · 金時玄]

I am a final-year Ph.D. student in Computer Science at Korea University, advised by Prof. Hyunwoo J. Kim. I also had the opportunity to spend part of my Ph.D. as a Visiting Scholar at the University of Wisconsin–Madison, working with Prof. Vikas Singh. I received my B.S. in Biomedical Engineering from Korea University.

My research seeks the essence beneath complex systems with machine learning: the representations that reveal hidden structures, and the dynamics that let us sample, transform, and create. Recently, I'm interested in guiding generative systems toward human, task, or reward-driven objectives, especially understanding where in the generative dynamics alignment can happen most effectively. Here is my CV.

Outside of research, I enjoy watching football, listening to alternative rock, and building custom keyboards. I have supported Liverpool F.C. since 2010.

Sihyeon Kim Sihyeon Kim getting an autograph from Sadio Mane

Research

My work sits at the intersection of 3D vision, geometric representation learning, and generative modeling. I have worked across images and videos, point clouds and graphs, explicit geometry and implicit neural fields. Across these domains, I am drawn to the systems behind the data, where structure explains what is observed and dynamics governs how states move. This leads to two main directions in my work: finding structure and learning dynamics.

Finding Structures

Representations that expose hidden structure in complex data, from local geometry and semantic parts to continuous neural fields and objective-aligned latent spaces.

Learning Dynamics

Methods for moving through data and distribution spaces, from controlled graph augmentation to acceleration-based flows and neural solvers for few-step sampling.

Publications

DDMI: Domain-Agnostic Latent Diffusion Models for Synthesizing High-Quality Implicit Neural Representations

Dogyun Park, Sihyeon Kim, Sojin Lee, Hyunwoo J. Kim

ICLR 2024 [arXiv] [code]

Generative Models Neural Fields Latent Spaces

Advancing Bayesian Optimization via Learning Correlated Latent Space

Seunghun Lee*, Jaewon Chu*, Sihyeon Kim*, Juyeon Ko, Hyunwoo J. Kim

NeurIPS 2023 [arXiv]

Bayesian Optimization Latent Spaces Regularization

Semantic-Aware Implicit Template Learning via Part Deformation Consistency

Sihyeon Kim, Minseok Joo, Jaewon Lee, Juyeon Ko, Juhan Cha, Hyunwoo J. Kim

ICCV 2023 [arXiv] [code]

Neural Fields Geometric Representation Shape Correspondence Regularization

Self-positioning Point-based Transformer for Point Cloud Understanding

Jinyoung Park*, Sanghyeok Lee*, Sihyeon Kim, Yunyang Xiong, Hyunwoo J. Kim

CVPR 2023 [arXiv] [code]

Point Clouds Geometric Representation Transformers

Point Cloud Augmentation with Weighted Local Transformations

Sihyeon Kim*, Sanghyeok Lee*, Dasol Hwang, Jaewon Lee, Seong Jae Hwang, Hyunwoo J. Kim

ICCV 2021 [arXiv] [code]

Point Clouds Geometric Representation Data Augmentation

ST-VLM: Kinematic Instruction Tuning for Spatio-Temporal Reasoning in Vision-Language Models

Dohwan Ko*, Sihyeon Kim*, Yumin Suh, Vijay Kumar B.G, Minseo Yoon, Manmohan Chandraker, Hyunwoo J. Kim

arXiv 2025 [arXiv] [project]

Video-LLM Spatio-Temporal Reasoning 4D Reconstruction

Constant Acceleration Flow

Dogyun Park, Sojin Lee, Sihyeon Kim, Taehoon Lee, Youngjoon Hong, Hyunwoo J. Kim

NeurIPS 2024 [arXiv] [code]

Generative Models Efficient Sampling

Metropolis-Hastings Data Augmentation for Graph Neural Networks

Hyeonjin Park*, Seunghun Lee*, Sihyeon Kim, Jinyoung Park, Jisu Jeong, Kyung-Min Kim, Jung-Woo Ha, Hyunwoo J. Kim

NeurIPS 2021 [arXiv] [code]

GNN Sampling Geometric Representation Data Augmentation

Learning to Solve Generative ODEs Beyond the Linear Span

Sihyeon Kim, Seunghun Lee, Vikas Singh§, Hyunwoo J. Kim§

arXiv 2026 [arXiv]

Generative Models Solver Learning Efficient Sampling