Our goal is to develop real unsupervised learning algorithms in the sense that they have real intelligence. We would like to call a family of these algorithms as self-evaluation algorithms.
We used to work on multichannel blind deconvolution and equalization problem which is very fundamental and challenging and has numerous applications in digital communication and wireless communications as well as in brain science. We also have a variety of interests in brain science, especially computational neuroscience and in a variety of topics in information theory, mainly information-theorectic learning.
Currently we are working on machine learning, especially statistical machine learning which included many exciting things such as probabilistic models, graphical models, kernel machines, Bayesian learning, and so on. We are also working on the applications of machine learning, which include brain computer interface, pattern classification, medical imaging, computational hearing/vision, bioinformatics, etc. We would like to keep doing some theoretical work to develop real learning algorithms and will do some application work as well.
Although whole algorithms can not be named as learning algorithms (they are simply adaptive algorithms, they are not really learning), we have called them as learning algorithms. Until we come up with self-evaluation algorithms, we will be excused to call them as learning algorithms.
News
 |
Four papers will be presented at ICASSP-2010 Following papers will be presented at the 34th International Conference on Acoustics, Speech, and Signal Processing (ICASSP-2010) See the Publication section for more details. |
 |
S. Kim has been completed his Ph.D. defense successfully. Sangki Kim has been completed the final defense of his Ph. D. dissertation. The thesis is about “Dynamic hand gesture recognition with aceeleromete”. He has developed dynamic hand gesture recognition system with aceelerometer. It includes isolated gesture recognition and continuous gesture recognition. Also, He suggest a new concept (pseudo velocity signals), which is transformed from acceleration signals, to improve signal’s discriminative power. |
 |
J. Kim has been completed his Ph.D. dissertation defense. Jong Kyung Kim has been completed his Ph. D. dissertation successfully. The thesis title is “Probabilistic models for motif discovery in biopolymer sequences.” He has developed probabilistic models which allow effective motif discovery and sequence classification In biopolymer sequences. |
 |
J.K. Kim is awarded TJ Park Bessemer Science Fellowship Jong Kyung Kim (Ph. D. Candidate) is awarded TJ Park Bessemer Science Fellowship (Post-doc position), which provides about 60,000 U.S. dollars for two years to young highly-intelligent researchers in Korea. TJ Park Bessemer Science Fellowship is first in this year and a competitive ratio is 8:1. |
 |
Y. Kang was awarded SAS Paper Award Yoonseop Kang was awarded SAS Paper Award (SAS 우수 학술상) on 2009 The Korean Datamining Society fall conference for his paper “Kernel PCA for Community Detection”. In the paper, The authors suggested a community detection method which uses kernel PCA with kernel function designed to represent similarity between network nodes. |