Jae-Won Chung

GPU cluster resource management

• Advised by Professor Byung-Gon Chun.
• Developing Crane, an elastic GPU cluster resource manager for deep learning workloads.
• Kubernetes backend for Crane, efficient AutoML scheduling for GPU clusters.

Server-client collaborative DNN inference

• Advised by Professor Soo-Mook Moon.
• Developed a novel server-client collaborative video DNN inference scheme that drastically reduces network traffic via intermittent knowledge distillation.
• Implemented the scheme with PyTorch & OpenMPI and conducted evaluation using an NVIDIA Jetson Nano board as a client

Meta-learning, Few-shot Classification

• Advised by Professor Kyoung Mu Lee.
• Conducted research on better meta-initialization points for Model-Agnostic Meta-Learning (MAML) using an LSTM-based neural memory.
• Conducted research on generating task-aware class embeddings to augment feature maps of MAML with a convex program (DPP).

Deep Learning methods for Quantitative Susceptibility Mapping (QSM)

• Advised by Professor Jongho Lee.
• Designed, implemented, and trained a U-Net variant on in-vivo brain MRI field images and their COSMOS results.
• Submitted our solution to the QSM challenge held by the 5th International Workshop on MRI Phase Contrast and QSM.

A free research group on all domains of deep learning

• Gained experience extensively in computer vision and meta-learning, and attended talks on computer vision, natural language processing, reinforcement learning, and speech recognition.
• Gave a talk with the title "Memory plus Meta-Learning".

An official Coursera community that traslates Coursera lecture subtitles

• Served as Language Coordinator, a selected position that reviews and confirms works by other translators.
• Created Korean subtitles for Coursera Lectures initially provided only in English. Focused on courses related to machine learning.