Jaewoo Baek

Jaewoo Baek

Auto-generated profile, based on publicly available information

Jaewoo Baek

Intelligence C4I Team, Hanwha Systems, Seoul, Gyeonggi-do, Republic of Korea

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
30 March 2022 in IEEE Access
Verified
Journal Article
Open access
JBJaewoo Baek
Jaewoo Baek
CLChoongseop Lee
Choongseop Lee
Hyunsoo YuHyunsoo Yu
Hyunsoo Yu
Suwhan BaekSuwhan Baek
Suwhan Baek
SLSeokmin Lee
Seokmin Lee
CPCheolsoo Park
Cheolsoo Park
Sleep experts manually label sleep stages via polysomnography (PSG) to diagnose sleep disorders. However, this process is time-consuming, requires a lot of labor from sleep experts, and makes the participants uncomfortable with the attachment of multiple sensors. Thus, automatic sleep scoring methods are essential for practical sleep monitoring in our daily lives. In this study, we propose an automatic sleep scoring model based on intrinsic oscillations in a single channel electroencephalogram (EEG) signal. We applied noise assisted bivariate empirical mode decomposition (NA-BEMD) to extract the intrinsic mode components and an attention mechanism in deep neural networks to provide weights to the components depending on their significance to sleep scoring. In particular, through the attention mechanism, we found an interpretable model by examining the oscillations that correspond to specific sleep stages. Therefore, we analyzed which frequency components are more weighted to a sleep stage than the others, when the model classifies sleep stages, and, as a result, confirmed that the model assigns convincing weights to the frequency components for each sleep stage. Additionally, the model consists of a one-dimensional convolutional neural network (1D-CNN) to extract features of an epoch and bidirectional long short-term memory (Bi-LSTM) to learn the sequential information of the consecutive epochs. We evaluated proposed model using Fpz-Cz, Pz-Oz, and F3-M2 channel EEG from three different public datasets (Sleep-EDF-2013, Sleep-EDF-2018, WSC) and demonstrated that our model yielded the best overall accuracy (Fpz-Cz: 86.22%-82.67%, Pz-Oz: 83.63%-80.15%, F3-M2: 84.20%) and macro F1-score (Fpz-Cz: 80.79%-76.90%, Pz-Oz: 76.89%-72.98%, F3-M2: 74.88%) compared with the state-of-the-art sleep scoring algorithms using single channel EEG. As a benchmark test, FIR bandpass filters were compared, and it was confirmed that NA-BEMD was superior to the traditional filters in all experiments, demonstrating that the proposed model is interpretable and a state-of-the-art sleep scoring algorithm.

21 February 2025 in IEEE Access
Verified
Journal Article
Open access
CLChoongseop Lee
Choongseop Lee
GYGeunbo Yang
Geunbo Yang
JBJaewoo Baek
Jaewoo Baek
YPYuntae Park
Yuntae Park
MCMingyu Cheon
Mingyu Cheon
CPCheolsoo Park
Cheolsoo Park
Spiking neural networks, known for mimicking the brain’s functionality resulting in efficient algorithms, are gaining attention across various problems and applications. However, their potential in regression tasks remains relatively unexplored. This study focuses on leveraging the spiking neural architecture in conjunction with Fourier analysis and support vector regression to estimate heart rates from electrocardiogram signal. We evaluated the regression errors of our model using three distinct elctrocardiogram datasets and assessed its performance on neuromorphic hardware by embedding spike-based layers. Our findings reveal that, compared to the conventional deep learning models, the proposed spiking neural system achieves a computational efficiency improvement while maintaining the competitive regression accuracy. Finally, we discuss the regression performance, energy efficiency, biological plausibility, and potential applications of the proposed neuromorphic system.