Research Overview

Research Statement

• Deep Learning (DL) has attracted increasing attentions, showing promising performance in many AI tasks being advanced rapidly. We, the researchers in Machine Learning and Visual Computing (MLVC) Lab, aim to pave the way of creating new DL architectures, algorithms and applications which are much more powerful, efficient and widely applicable. Our efforts will push the boundaries of human knowledge and improve the quality and safety of life through AI technology. The specific research topics in MLVC Lab are as follows:

1. Data efficiency

• Data augmentation [IEEE Access 2021], [ICLR 2021]

• Semi-supervised learning [IEEE Access 2021]

2. Model efficiency and automation

• Distillation [ICCV 2021], [ECCV 2022]

• Pruning [IEEE Access 2021]

• Quantization [IEEE Access 2021a, 2021b]

• Filter decomposition [IEEE Access 2021, 2022]

• Neural architecture search  [IEEE Access 2021]

• Neuromorphic system [IEEE Access 2020]

3. Robustness and explainable AI  

• Robustness [IEEE TNNLS 2020], [ICCV2021], [NeurIPS 2021], [NeurIPS 2022]

• Explainable AI [IEEE Access 2022]

4. AI+X

• Image restoration [IEEE Access 2019a, 2019b, 2020] 

• Semantic segmentation [Info. Sci. 2020] 

• Facial expression [IEEE TAC 2020]

• Video compression [IEEE TIP 2018]

• Material science [2D Materials 2021]

• Astronomy [Nature Ast. 2019], [AJL 2020] 

• Medical science [IEEE IJMI 2019], [IEEE JBHI 2020]

Collaborators

(in alphabetical order)

• Prof. Tae-Choong Chung (Kyung Hee University, South Korea)

• Dr. Soo Ye Kim (Adobe Research, USA)

• Prof. Phillip Kim (Harvard University, USA)

• Prof. Jong Hwan Ko (Sungkyunkwan University, South Korea)

• Prof. Sungyong Lee (Kyung Hee University, South Korea)

• Prof. Yong-Jae Moon (Kyung Hee University, South Korea)

• Prof. Tae-Hyun Oh (POSTECH, South Korea)

• Dr. Young Jae Shin (Brookhaven National Lab, USA)

• Prof. Simon Woo (Sungkyunkwan University, South Korea)