Dongsu Song

About Me

Open to opportunities

Seeking a Ph.D. position

Adversarial Machine Learning · Robust Computer Vision

Hello! I am Dongsu Song. Thank you for your interest. I received my M.S. degree in Artificial Intelligence from Korea Aerospace University in 2025, and I was fortunate to conduct my master’s research under the supervision of Prof. Jay Hoon Jung. My research interests center on Adversarial Machine Learning. In particular, I am interested in stress-testing the robustness of computer vision models and developing methods to improve their robustness. I am also interested in using generative models to address safety-critical real-world scenarios that are difficult to learn from due to data scarcity, including long-tail cases.

My interests

Adversarial Machine Learning: Stress-testing computer vision models and developing methods to improve their robustness
Generative Models for Data-Scarce Scenarios: Addressing safety-critical, long-tail problems using generative models

News

[2026.06.18] Paper accepted at ECCV 2026!
[2026.06.02] Selected as a Reviewer for NLDL 2027.
[2025.08] Invited presentation at KCCV 2025!
[2025.06] Selected as a Reviewer for NeurIPS 2025.
[2024.11] Paper accepted at NeurIPS 2024!
[2023.10] Paper accepted at Journal of KIISE!

Selected Papers

SegPAR: Class-Centric Decision-Based Sparse Attack for Semantic Segmentation Selected

Dongsu Song, Daeyun Go, Boseung Seo, Jay Hoon Jung

European Conference on Computer Vision (ECCV), 2026 (Poster)

[Paper]

Amnesia as a Catalyst for Enhancing Black Box Pixel Attacks in Image Classification and Object Detection Selected

Dongsu Song, Daehwa Ko, Jay Hoon Jung

Conference on Neural Information Processing Systems (NeurIPS), 2024

[Paper] [Code]

Professional Activities

Academic Service
- NLDL 2027 Reviewer
- NeurIPS 2025 Reviewer

Invited Talk
- KCCV(Korean Conference on Computer Vision) 2025: "Amnesia as a Catalyst for Enhancing Black Box Pixel Attacks in Image Classification and Object Detection"