Scholar
Jaemin Kim
Google Scholar ID: QWRf4vgAAAAJ
KAIST AI
Citations & Impact
Citations
12
H-index
3
i10-index
0
Publications
6
Co-authors
8
list available
Publications
Resume (English only)
Academic Achievements
- Published papers include: 'Training-Free Reward-Guided Image Editing via Trajectory Optimal Control', 'Free2Guide: Training-Free Text-to-Video Alignment using Image LVLM', 'Universal Reasoner: A Single, Composable Plug-and-Play Reasoner for Frozen LLMs', 'Optical-Flow Guided Prompt Optimization for Coherent Video Generation', 'Derivative-Free Diffusion Manifold-Constrained Gradient for Unified XAI', 'Generalized Consistency Trajectory Models for Image Manipulation', 'HiCBridge: Resolution Enhancement of Hi-C Data Using Direct Diffusion Bridge'.
Research Experience
- Research focuses include connecting arbitrary distributions with Diffusion Models (DMs), generalizing Consistency Trajectory Models (CTMs) to connect arbitrary distributions, generating more natural motion through guidance in Video DMs, providing meaningful directions from DMs samples, using DMs samples as guidance, serving LLMs as guidance in RL approaches, and editing images to maximize reward in a training-free manner using DMs.
Education
- Pursuing Ph.D. at KAIST BISPL under the supervision of Prof. Jong Chul Ye; completed master’s degree at KAIST, also supervised by Prof. Jong Chul Ye.
Background
- Ph.D. student at KAIST Graduate School of AI, with a focus on controllable generative AI, particularly exploring ways to efficiently steer the outputs of advanced models such as diffusion models and large language models (LLMs). His work lies at the intersection of generative modeling, trajectory-level optimization, and black-box guidance, aiming to make generative AI more controllable and interpretable.
