Research Interest — P: preprint, J: journal, C: conference/workshop, A: abstract
I am currently interested in neural network compression (e.g., pruning [C15, C14, C12, P3, C11, C10, C9], knowledge distillation [C15, C14, C13, C11, C10], quantization [C13, C10]).
My research interests are in the areas of machine learning to process:
Image/Video: text-to-image generation [C15, C14, C13, C11], talking face generation [C10, C9], RGB-NIR video fusion [J8], controllable image synthesis [C9, Thesis, P1, J6], fashion image manipulation [J6, A8], facial expression recognition [J4, C7, C6, A7, Challenge Winner], object recognition [C4]
Text: language modeling [C12, P3, J7], dialogue act classification [C8], document classification [J1]
Biosignal: rPPG-based heart rate estimation [J8], EEG classification [J3, C2, A1] & analysis [C5, A6, A4], fMRI analysis [J2]
Speech: speech enhancement [J5], emotional sound recognition [C3], intention recognition [A3]
Publication — IF: impact factor (JCR 2022 release)
2024
[C15] BK-SDM: A Lightweight, Fast, and Cheap Version of Stable Diffusion [PAPER] [CODE]
Bo-Kyeong Kim, Hyoung-Kyu Song, Thibault Castells, Shinkook Choi
European Conference on Computer Vision (ECCV), 2024
[P3] Shortened LLaMA: Depth Pruning for Large Language Models with Comparison of Retraining Methods [PAPER] [CODE]
Bo-Kyeong Kim, Geonmin Kim, Tae-Ho Kim, Thibault Castells, Shinkook Choi, Junho Shin, Hyoung-Kyu Song
ArXiv, 2024
[C14] LD-Pruner: Efficient Pruning of Latent Diffusion Models using Task-Agnostic Insights [PAPER]
Thibault Castells, Hyoung-Kyu Song, Bo-Kyeong Kim, Shinkook Choi
CVPR Workshop on Efficient and On-Device Generation (EDGE), 2024
[C13] EdgeFusion: On-Device Text-to-Image Generation [PAPER]
Thibault Castells, Hyoung-Kyu Song, Tairen Piao, Shinkook Choi, Bo-Kyeong Kim, Hanyoung Yim, Changgwun Lee, Jae Gon Kim, Tae-Ho Kim
CVPR Workshop on Efficient and On-Device Generation (EDGE), 2024 — AK Daily Papers
[C12] Shortened LLaMA: A Simple Depth Pruning for Large Language Models [PAPER]
Bo-Kyeong Kim, Geonmin Kim, Tae-Ho Kim, Thibault Castells, Shinkook Choi, Junho Shin, Hyoung-Kyu Song
ICLR Workshop on Mathematical & Empirical Understanding of Foundation Models (ME-FoMo), 2024 — AK Daily Papers
2023
[C11] BK-SDM: Architecturally Compressed Stable Diffusion for Efficient Text-to-Image Generation [PAPER] [DEMO] [POSTER]
Bo-Kyeong Kim, Hyoung-Kyu Song, Thibault Castells, Shinkook Choi
ICML Workshop on Efficient Systems for Foundation Models (ES-FoMo), 2023 — AK Daily Papers
ICCV Demo Track, 2023 — title: Architecturally Compressed Stable Diffusion for Efficient Text-to-Image Generation [LINK]
[C10] A Unified Compression Framework for Efficient Speech-Driven Talking-Face Generation [PAPER] [DEMO] [SLIDE] [POSTER]
Bo-Kyeong Kim, Jaemin Kang, Daeun Seo, Hancheol Park, Shinkook Choi, Hyoung-Kyu Song, Hyungshin Kim, Sungsu Lim
MLSys Workshop on On-Device Intelligence (ODIW), 2023
ICCV Demo Track, 2023 — title: Accelerating Speech-Driven Talking Face Generation with 28× Compressed Wav2Lip [LINK]
NVIDIA GPU Technology Conference (GTC), 2023 — Poster spotlight [POSTER] [SLIDE]
2022
[J8] Self-supervised RGB-NIR Fusion Video Vision Transformer Framework for rPPG Estimation [PAPER]
Soyeon Park, Bo-Kyeong Kim, Suh-Yeon Dong
IEEE Transactions on Instrumentation & Measurement (TIM), 2022 — IF 5.332
[C9] Cut Inner Layers: A Structured Pruning Strategy for Efficient U-Net GANs [PAPER] [SLIDE] [POSTER] [VIDEO]
Bo-Kyeong Kim, Shinkook Choi, Hancheol Park
ICML Workshop on Hardware Aware Efficient Training (HAET), 2022
2021
[J7] Unsupervised Multi-Sense Language Models for Natural Language Processing Tasks [PAPER] [CODE]
Jihyeon Roh, Sungjin Park, Bo-Kyeong Kim, Sang-Hoon Oh, Soo-Young Lee
Neural Networks, 2021 — IF 9.657
2020
[P1] Semi-supervised Disentanglement with Independent Vector Variational Autoencoders [PAPER] [CODE]
Bo-Kyeong Kim, Sungjin Park, Geonmin Kim, Soo-Young Lee
ArXiv, 2020
2019
[Thesis] Learning-Based Image Synthesis with Disentangled Representations [THESIS] [SLIDE: PhD Defense] [SLIDE: Grad Research Portfolio]
Bo-Kyeong Kim
PhD Thesis, KAIST, Dec 2019
[J6] Style-Controlled Synthesis of Clothing Segments for Fashion Image Manipulation [PAPER] [CODE]
Bo-Kyeong Kim, Geonmin Kim, Soo-Young Lee
IEEE Transactions on Multimedia (TMM), 2019 — IF 8.182
[J5] Unpaired Speech Enhancement by Acoustic and Adversarial Supervision for Speech Recognition [PAPER] [CODE]
Geonmin Kim, Hwaran Lee, Bo-Kyeong Kim, Sang-Hoon Oh, Soo-Young Lee
IEEE Signal Processing Letters (SPL), 2019 — IF 3.201
[A8] Feature Disentanglement and Top-Down Attention for Emotional Conversational Agents [LINK]
Tae-Ho Kim, Bo-Kyeong Kim, Young Hoon Shin, Soo-Young Lee
Japan-China-Korea Joint Workshop on Neurobiology and Neuroinformatics (NBNI), 2019 — Oral presentation
2017
[C8] Compositional Sentence Representation from Character within Large Context Text [PAPER] [CODE]
Geonmin Kim, Hwaran Lee, Bo-Kyeong Kim, Soo-Young Lee
International Conference on Neural Information Processing (ICONIP), 2017 — Oral presentation
2016
[C7] Fusing Aligned and Non-Aligned Face Information for Automatic Affect Recognition in the Wild: A Deep Learning Approach [PAPER]
Bo-Kyeong Kim, Suh-Yeon Dong, Jihyeon Roh, Geonmin Kim, Soo-Young Lee
CVPR Workshop on Context-Based Affect Recognition and Affective Face in-the-Wild (CBAR & Wild), 2016 — Oral presentation
[J4] Hierarchical Committee of Deep Convolutional Neural Networks for Robust Facial Expression Recognition [PAPER] [CODE] [DEMO] [MEDIA]
Bo-Kyeong Kim, Jihyeon Roh, Suh-Yeon Dong, Soo-Young Lee
Journal on Multimodal User Interfaces (JMUI), 2016 — IF 1.694; Extension of [C6]
[J3] EEG-based Classification of Implicit Intention during Self-relevant Sentence Reading [PAPER] [DEMO]
Suh-Yeon Dong, Bo-Kyeong Kim, Soo-Young Lee
IEEE Transactions on Cybernetics (TCYB), 2016 — IF 19.118; Extension of [C2]
[J2] Implicit Agreeing/Disagreeing Intention while Reading Self-relevant Sentences: A Human fMRI Study [PAPER]
Suh-Yeon Dong, Bo-Kyeong Kim, Soo-Young Lee
Social Neuroscience, 2016 — IF 2.381
[A7] Hierarchical Committee and Top-Down Attention for Robust Classification: Cases for Emotional Facial Expression and Noisy Speech Recognition [LINK]
Bo-Kyeong Kim, Ho-Gyeong Kim, Soo-Young Lee
Japan-China-Korea Joint Workshop on Neurobiology and Neuroinformatics (NBNI), 2016 — Oral presentation
[A6] An EEG Study on Human Trust in Machine [LINK]
Suh-Yeon Dong, Bo-Kyeong Kim, Kyeongho Lee, Soo-Young Lee
Annual Meeting of the Organization for Human Brain Mapping (OHBM), 2016
2015
[C6] Hierarchical Committee of Deep CNNs with Exponentially-Weighted Decision Fusion for Static Facial Expression Recognition [PAPER] [CODE]
Bo-Kyeong Kim, Hwaran Lee, Jihyeon Roh, Soo-Young Lee
ACM International Conference on Multimodal Interaction (ICMI), 2015 — EmotiW'15 winner; Oral presentation
[J1] Hierarchical Feature Extraction by Multi-layer Non-negative Matrix Factorization Network for Classification Task [PAPER]
Hyun Ah Song, Bo-Kyeong Kim, Thanh Xuan Luong, Soo-Young Lee
Neurocomputing, 2015 — IF 5.779
[C5] A Preliminary Study on Human Trust Measurements by EEG for Human-Machine Interactions [PAPER]
Suh-Yeon Dong, Bo-Kyeong Kim, Kyeongho Lee, Soo-Young Lee
International Conference on Human-Agent Interaction (HAI), 2015
[A5] Emergent Situation Classification with Environmental Sound [LINK]
Taehwan Kwon, Jiyoung Kim, Bo-Kyeong Kim, Dong-Gun Lee, Soo-Young Lee
Japan-China-Korea Joint Workshop on Neurobiology and Neuroinformatics (NBNI), 2015
[A4] Electrophysiological Correlates of Reasoning during a Theory-of-mind Game [LINK]
Suh-Yeon Dong, Bo-Kyeong Kim, Kyeongho Lee, Soo-Young Lee
Japan-China-Korea Joint Workshop on Neurobiology and Neuroinformatics (NBNI), 2015
2014
[C4] Color Image Processing based on Nonnegative Matrix Factorization with Convolutional Neural Network [PAPER]
Thanh Xuan Luong, Bo-Kyeong Kim, Soo-Young Lee
International Joint Conference on Neural Networks (IJCNN), 2014
[A3] Understanding Shopper’s True Intention from Conversational Dialogue between Salesperson and Shopper [LINK]
Soo-Young Lee, Bo-Kyeong Kim
Annual Meeting of the Cognitive Science Society (COGSCI), 2014
2013
[C3] Spectral Feature Extraction Using dNMF for Emotion Recognition in Vowel Sounds [PAPER]
Bo-Kyeong Kim, Soo-Young Lee
International Conference on Neural Information Processing (ICONIP), 2013 — Oral presentation
[C2] Decoding and Predicting Implicit Agreeing/Disagreeing Intention Based on Electroencephalography (EEG) [PAPER]
Suh-Yeon Dong, Bo-Kyeong Kim, Soo-Young Lee
International Conference on Neural Information Processing (ICONIP), 2013 — Oral presentation
2012
[C1] Trust and Equity Theory in Prisoner's Dilemma [PAPER]
Eun-Soo Jung, Bo-Kyeong Kim, Soo-Young Lee
International Conference on Neural Information Processing (ICONIP), 2012
[A2] Social Interaction between Intelligent Individuals and Spontaneous Cooperation [LINK]
Eun-Soo Jung, Bo-Kyeong Kim, Soo-Young Lee
Japan-China-Korea Joint Workshop on Neurobiology and Neuroinformatics (NBNI), 2012
[A1] Discriminant Feature Extraction Based on Nonnegative Matrix Factorization for BCI System [LINK]
Bo-Kyeong Kim, Suh-Yeon Dong, Hyun Ah Song, Soo-Young Lee
International Joint Conference on Neural Networks (IJCNN), 2012