Research: Large-scale Deep Learning

In recent years, deep learning models such as CNN and DBN have emerged as a new and powerful paradigm for pattern recognition and classification. We are doing research on a scale-out deep learning system exploiting the computing power of GPUs. We are also interested in developing a deep interpretation model of multi-channel signals, especially invasive brain signal such ECoG. Currently, we focus on interpretation of real-time 32-channel signals through medical experiments using mouse as animal model.

Related Projects

Development of Intelligent Interaction Technology Based on Context Awareness and Human Intention Understanding

IITP AI Flagship, Ministry of Science, Korea
Dec. 2016 ~ Aug. 2021

BigLearning: A Large-scale Optimization System of Deep Neural Network Models (PI)

Samsung Research Funding Center, Samsung Electronics, Korea
Dec. 2015 ~ Nov. 2018

Rehabilitation and replacement technology for brain damage employing electrical method

DGIST, Ministry of Science, Korea
Jan. 2012 ~ Dec. 2016

Related Papers

Coming soon…

Patents

Robot avatar system using hybrid interface and command server and sensory server therefor

Kim, M.-S., Son, S. H., Kim, J., Jeon, K., and Hong, I.
Korean Patent, Appl. No. 2013-0000646, Jan. 3, 2013, Reg. No. 10-1343860, Dec. 16, 2013.