Geon Yeong Park

About Me

I am a third-year Ph.D. student in Bio and Brain Engineering (BBE) at KAIST, with a strong desire to become a tech-based leader in AI. Fortunately co-advised by Prof. Jong Chul Ye and Prof. Sang Wan Lee. Completed M.S. in BBE at KAIST, advised by Prof. Sang Wan Lee.

I am especially interested in (a) improving the learning of representations, and (b) leveraging these intermediate representations for controlling AI systems. My recent research focuses on controlling the diffusion generative process by exploiting latent representations or denoised estimates, with the goal of improving the conditional diffusion sampling and multi-modal alignment. Additionally, I am interested in developing robust representations to handle various distribution shifts.

Research Interests

• Generative models: Diffusion models and their applications. (Editing, Inverse imaging, etc)
• Multi-modal learning: Text, Image, Video, 3D, etc.
• Robust representation learning: Adversarial vulnerability, Debiasing, Transfer learning.

News

• [Jul. 2024] Two papers (Dreamsampler, DreamMotion) accepted at ECCV 2024.
• [Jun. 2024] CFG++ is available on arxiv. A simple fix to CFG that enables lower guidance scales, improves sample quality and invertibility.
• [Mar. 2024] Dreamsampler, DreamMotion, Treg is available on arxiv.
• [Mar. 2024] Three papers (Self-supervised debiasing, VMC, CDS) accepted at CVPR 2024.
• [Jan. 2024] EBCA receives 30th Samsung Humantech Paper Award (3rd in Signal Processing)
• [Dec. 2023] One poster session (EBCA) at NeurIPS 2023 in New Orleans.

Publications

Services

Conference Reviewers

• IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2024
• European Conference on Computer Vision (ECCV) 2024

Journal Reviewers

• IEEE Transactions on Image Processing (TIP)

Powered by Jekyll and Minimal Light theme.