Researcher working across machine learning theory and practical AI systems
Researcher in Artificial Intelligence
Korea Aerospace University
Google Scholar · OpenReview · ORCID · Email
AISTATS 2026 PMLR 300 Learnable Activations
A learnable sinusoidal activation that provides explicit control over polynomial degree and selective amplification of high-order components.
Daehwa Ko, JaeHyeon Kim, SeungHyun Ham, Jay Hoon Jung
Proceedings of the 29th International Conference on Artificial Intelligence and Statistics (AISTATS 2026), PMLR 300
Read the paper → · OpenReview →
Amnesia as a Catalyst for Enhancing Black Box Pixel Attacks in Image Classification and Object Detection
NeurIPS 2024 Adversarial Robustness Computer Vision
Dongsu Song, Daehwa Ko, Jay Hoon Jung
Advances in Neural Information Processing Systems (NeurIPS 2024)
- Title withheld for double-blind review — Submitted to NeurIPS 2026
- Title withheld for double-blind review — Submitted to AAAI 2026
- Manuscript I — title withheld for double-blind review — Submitted to ICLR 2027
- Manuscript II — title withheld for double-blind review — Submitted to ICLR 2027
Additional Publications
Daehwa Ko, Jay Hoon Jung
Korean Society for Aeronautical & Space Sciences Spring Conference, 2025
Outstanding Presentation Paper Award
Daehwa Ko, JaeHyeon Kim, Jay Hoon Jung
KICS Summer Conference, 2024
VoiceKey: Real-Time Compression Encoding and Quantization for Voice Authentication Models on Edge Devices
Daehwa Ko, Garam Jung, Jay Hoon Jung
The 4th Korea Artificial Intelligence Conference
September 2024 – Present
Building an applied generative-AI product from model development to production deployment.
- Leading the development of GlowMe, an AI-powered beauty and style transformation application.
- Serving as CEO and lead AI developer.
- Developing and operating the product across Android and iOS.
- Building applied generative-AI features for personalized hairstyle, makeup, and style previews.
- Received the Excellence Award at the SNU SNaaC Re:Pit Acceleration Program.
- Recognized at Goyang IR Day for business potential, marketability, and technical innovation.
Website → · Google Play → · App Store →
December 2021 – Present
- Conducting research in machine learning, reinforcement learning, adversarial robustness, and computer vision.
- Developed reinforcement-learning methods for autonomous navigation in colonoscopy procedures.
- Constructed and optimized simulation environments for training and evaluating reinforcement-learning agents.
- Led research activities for the AIER industry–academia project involving Korea Aerospace University, Korean Air, and Airbus.
- Investigated aviation-AI robustness and model vulnerabilities using explainable-AI techniques.
- Authored educational materials covering reinforcement learning, classification, object detection, and representation learning.
- Developed explanations connecting theoretical concepts with practical implementations and code examples.
- Collaborated with other contributors to improve the accuracy and accessibility of the material.
Reinforcement Learning → · Classification → · Object Detection → · Representation Learning →
Additional Experience, Leadership, and Activities
March 2024 – September 2025
- Delivered hands-on AI education programs for young learners.
- Introduced practical machine-learning concepts through interactive experiments and projects.
April 2023 – December 2023
- Led a nonprofit organization focused on socially beneficial AI applications.
- Developed a domestic-violence prevention application in collaboration with Ilsan East Police Station.
- Developed an AI-based speaker-verification system for domestic-violence prevention and response.
- Participated in the Goyang City business-funding program.
- Published the VoiceKey research project at the 4th Korea Artificial Intelligence Conference.
2023 – Present
- Lead the deep-learning club at Korea Aerospace University.
- Organize machine-learning projects and technical study sessions.
- Mentor new members and support collaborative research and development.
- Initiated and managed multiple deep-learning projects across the club.
August 2022 – December 2022
- Collaborated with Ratio on a computer-vision system for autonomous driving.
- Developed vision-AI components for recognizing road environments and driving-related objects.
- Developed a web application integrating a counseling AI system.
- Contributed to the project receiving second place in the associated industry–academia program.
- Operate the YouTube channel Daehwa's AI & Linear Algebra Deep Dive →.
- Produce lectures on deep learning, linear algebra, and the mathematical foundations of artificial intelligence.
| Year | Recognition |
|---|---|
| 2025 | Outstanding Presentation Paper Award, Korean Society for Aeronautical & Space Sciences Spring Conference |
| 2025 | Excellence Award, SNU SNaaC Re:Pit Acceleration Program |
| 2025 | Goyang IR Day Award, recognized for business model, market potential, and technical innovation |
| 2024 | Grand Prize, KAU AI Drone Racing |
| 2024 | Grand Award, KAU Generative AI Business Hackathon |
Undergraduate Awards and Projects
- Served as team leader for the Really You project.
- Led planning and backend development.
- Extended a lightweight deepfake-detection concept to address deepfake voice generation.
- Developed a startup service based on lightweight deepfake detection.
- Applied controlled image distortions to reduce the risk of images being reused for deepfake generation.
- Developed a deep-learning framework for real-time information validation and improved communication security.
- Developed an image-recognition system for autonomous driving.
- Focused on Korean traffic signs and locally collected road-scene data.
- Developed a serverless backend architecture.
- Integrated KoBERT into an AI-powered counseling chatbot.
- Developed a Unity-based system for generating synthetic images for machine-learning model training.
M.S. in Artificial Intelligence
Department of Artificial Intelligence
March 2024 – Present
B.S. in Software and Mechanical Engineering
Double Major
March 2020 – February 2024