Research‎ > ‎

Algorithm/Methodology developments

Algorithm/Methodology developments
(1) Improvement of classification performance via deep neural network
Deep neural network (DNN) with weight sparsity control (i.e., L1-norm regularization) improved the classification performance using whole-brain resting-state functional connectivity pattern of schizophrenia patient and healthy groups. Initializing DNN's weights through stacked auto-encoder enhanced the classification performance as well. (Kim et al., NI, 2016)





(2) Performance enhancement of independent component analysis (ICA) from group fMRI data
An iterative dual-regression (iDR) approach with a sparse prior regularization showed the better performance to estimate neuronal activation in the individual level using the result of an ICA. e.g. Iterative approach of dual regression with a sparse prior enhances the performance of independent component analysis for group fMRI data (Kim et al., NI, 2012)
 

       
   
 
       
(3) Machine learning techniques to fMRI data
Application of the artificial neural networks (ANN) techniques and machine learning algorithms to the fMRI. e.g. Independent vector analysis (IVA) to group fMRI data (Lee et al., NI, 2008)
 

                                          

         The 3-dimensional IVA model to fMRI                                                                 IVA result using the motor task