Ph.D. Candidate, Multi-Modal Artificial Intelligence Lab, Ajou UniversityAdvised by Prof. Jongbin Ryu
I make large neural networks efficient enough to run under tight compute limits, on edge devices and inside robot control loops.
Paper accepted to NeurIPS 2026.
Two papers accepted to EMNLP 2026.
Paper accepted to BMVC 2026.
Selected for the Doctoral Research Grant from NRF Korea.
First place at the IEEE Low-Power Computer Vision Challenge, CVPR 2025.
Paper accepted to ACCV 2024.
Filed a PCT patent on CT-based small bowel obstruction triage.
Paper published in Scientific Reports.
Paper published in the International Journal of Surgery.
I want VLA policies to hold up inside real robot control loops, which means three things at once: training that transfers across embodiments, inference cheap enough to run every control step, and failures caught early enough for the policy to recover.
Structured pruning leaves a model that has to relearn what was removed. I work on giving that recovery enough capacity, and on scheduling compression so the error that accumulates in deeper layers is paid down first (EMNLP 2026).
Making vision models cheap without giving up accuracy: re-parameterization at branch level (ACCV 2024), transfer that keeps a vision-language model general (BMVC 2026), and test-time adaptation that verifies candidate labels instead of adjusting predictions (NeurIPS 2026). The same drive won Track 1 of the 2025 IEEE Low-Power Computer Vision Challenge.
Neural Information Processing Systems (NeurIPS), 2026
Empirical Methods in Natural Language Processing (EMNLP), 2026
Empirical Methods in Natural Language Processing (EMNLP), 2026
British Machine Vision Conference (BMVC), 2026
arXiv preprint, 2026
Asian Conference on Computer Vision (ACCV), 2024
Scientific Reports, 2024
International Journal of Surgery, 2023 (IF 15.3, JCR top 1%)
Jongbin Ryu, Dooyeon Na, Seungmin Oh, Donggeon Lee
Korean Patent Application No. 10-2026-0123632filed Jul 2026
Hojung Shin, Jongbin Ryu, Seungmin Oh
PCT International Patent Application No. PCT/KR2024/014604filed Sep 2024
Real-time detection transformer for autonomous driving
Source-free domain adaptation for on-device face recognition
Masked-face recognition with domain adaptation
Lymphoma recognition and survival analysis from 3D CT
NRF KoreaSep 2025 - Aug 2027
Progressive Linear Activation for Depth-wise Reparameterization of Large-scale Models
NIA KoreaOct 2022 - Dec 2023
Track 1 of 45 teams, at the CVPR workshop2025
Layer Imitation: Efficient Vision-Language Model on the Mobile NPU
Google2023
Smiler: An Expression-Training App for Children with ASD
ACK, KOSIA2021
Recommender System for Web Search Based on NLP to Improve User Search Environment
Korea Digital Contents Society2021
Lightweight Video Search Using Video-to-Text Generation and Retrieval
Changing what the policy is trained to predict, so skills carry across robots instead of binding to one.
Cutting the compute spent on every control step while keeping task success.
Noticing that an action is going wrong early enough for the policy to recover.
Letting regularization respond to the data instead of applying one fixed strength.
I built and operated our lab's GPU cluster, which consists of about 20 nodes with eight RTX 3090s each. This work included containerized provisioning, per-node CUDA isolation, and training jobs that ran unattended for days. Outside of research, I taught AI tutorials for undergraduate research assistants and helped launch the Google Developer Student Club chapter at Ajou University.