Seongha Eom

I am a MS student at OSI Lab(advised by Prof. Se-young Yun) in KAIST AI. My research interests are in computer vision, transfer learning, and representation learning. Currently, my research topic is an efficient transfer learning method in multi-label classification. I try to achieve both efficiency and performance which is challenging but interesting. I believe considering the trade-offs helps solving practical problems of applying ML/DL models in various real-world tasks.

Layover Intermediate Layer for Multi-Label Classification in Efficient Transfer Learning

This paper achieves efficient transfer learning by utilizing intermediate representation and feature extracted from pre-trained network for multi-label classification

Real-time and Explainable Detection of Epidemics with Global News Data

This paper achieves early detection of epidemics, including COVID-19, by processing real-time global news data into graph clusters and retrieving explainable information with efficacy

Detection and prediction model for infectious disease

Supported by Institute for Security Convergence Research (ISCR), Developed detection model for epidemic prevention framework , Feb 2022 – Dec 2022

Server Manager (Czar)

Supported for setting environments and managing resources for shared lab servers, November 2021 - August 2022

초거대 인공지능 API 의료 분야 적용 아이디어 경진대회 (6위)

Hosted by NAVER Cloud, HyperCLOVA, Sept 2022

2022 Neurips Weather4cast Challenge (4th place)

Hosted by IARAI(Institute of Advanced Research in artificial Intelligence), just accepted at Weather4cast 2022 Workshop at the NeurIPS 2022 Competition Track!