(* indicates equal contribution)
GAN Training
This research introduces a novel GAN training methodology, Selective Focusing Learning (SFL), which enforces the discriminator and generator to learn the easy samples rapidly while maintaining diversity.
•
Kyeongbo Kong*, Kyunghun Kim*, Woo-Jin Song, Suk-Ju Kang, "Selective Focusing Learning for Conditional GANs", ICMLw, 2021. (Top-tier AI conference workshop) (Spotlight)
•
Kyeongbo Kong, Kyunghun Kim, Suk-Ju Kang, "Robust cGAN Training with Selective Matching", (Submitted)
Noisy Label
In general, deep neural network is vulnerable to noisy labels also known as erroneous labels. We propose robust learning methodology for noisy labels.
•
Kyeongbo Kong, Junggi Lee, Youngchul Kwak, Minsung Kang, Seong Gyun Kim, Woo-Jin Song, "Recycling: Semi-Supervised Learning With Noisy Labels in Deep Neural Networks", IEEE Access, 2019.
•
Junggi Lee, Kyeongbo Kong, Woo-Jin Song, "Learning algorithm with noisy labels for video frame classification", Journal of the Society for Information Display, 2020.
•
Kyeongbo Kong*, Junggi Lee*, Youngchul Kwak, Young-Rae Cho, Seong-Eun Kim, Woo-Jin Song, "Mitigating Memorization in Sample Selection for Learning with Noisy Labels ", ICMLw, 2021. (Top-tier AI conference workshop) (Spotlight)
•
Kyeongbo Kong, Junggi Lee, Youngchul Kwak, Young-Rae Cho, Seong-Eun Kim, Woo-Jin Song, "Penalty based Robust Learning with Noisy Labels", Neurocomputing, 2022. (IF: 5.719)
Instance Selection
This research proposes a data curation methodology via instance selection to reduce the search time of Neural Architecture Search (NAS).
•
Jae-hun Shim*, Kyeongbo Kong*, Suk-Ju Kang, "Core-set Sampling for Efficient Neural Architecture Search", ICMLw, 2021. (Top-tier AI conference workshop) (Spotlight)
•
Jae-hun Shim*, Kyeongbo Kong*, Suk-Ju Kang, "Accelerating Neural Architecture Search via Sample Selection", (Submitted)
Image Clustering
We propose a novel conversion framework from single modality to multi modality for image clustering, which improves the image clustering performance using text.
•
Daehyeon Kong*, Kyeongbo Kong*, Suk-Ju Kang, "Image Clustering using Generated Text Centroids", (Submitted)
Communications
We propose a deep learning-based channel estimation, quantization, feedback, and precoding method for downlink multiuser multiple-input and multiple-output systems.
•
Kyeongbo Kong, Woo-Jin Song, Moonsik Min, "Knowledge Distillation-aided End-to-End Learning for Linear Precoding in Multiuser MIMO Downlink Systems with Finite-Rate Feedback", IEEE Transactions on Vehicular Technology, 2021. (IF: 5.978)
Brain Wave
Mental workload is defined as the proportion of the information processing capability used to perform a task. We propose a three-dimensional convolutional neural network (3D CNN) employing a multilevel feature fusion algorithm for mental workload estimation using electroencephalogram (EEG) signals.
•
Youngchul Kwak, Kyeongbo Kong, Woo-Jin Song, Byoung-Kyong Min, Seong-Eun Kim, "Multilevel Feature Fusion With 3D Convolutional Neural Network for EEG-Based Workload Estimation", IEEE Access, 2020.
•
Youngchul Kwak, Kyeongbo Kong, Woo-Jin Song, Seong-Eun Kim, "Subject-Invariant Deep Neural Networks Based on Baseline Correction for Brain-Computer Interfaces", IEEE Journal of Biomedical and Health Informatics, 2023. (IF: 7.021)
SAR/IR
We propose dissimilarity regularization with a multistage fusion stream for a synthetic aperture radar (SAR) and infrared (IR) sensor fusion using deep learning.
•
Young-Rae Cho, Seungjun Shin, Sung-Hyuk Yim, Kyeongbo Kong, Hyun-Woong Cho, Woo-Jin Song, "Multistage Fusion With Dissimilarity Regularization for SAR/IR Target Recognition", IEEE ACCESS, 2019.