Multi-Voxel Pattern Analysis of fMRI Based on Deep Learning Methods
A decoding process for fMRI data is constructed based on Multi-Voxel Pattern Analysis (MVPA) using deep learning method for online training process. The constructed process with Deep Brief Network (DBN) extracts the feature for classification on each ROI of input fMRI data. The decoding experiment results for hand motion show that the decoding accuracy based on DBN is comparable to that with the conventional process with batch training and that the divided feature extraction in the first layer decreases computational time without loss of accuracy. The constructed process should be necessary for interactive decoding experiments for each subject.
KeywordsfMRI MVPA deep learning Deep Brief Network
Unable to display preview. Download preview PDF.
- 4.Bengio, Y., Lamblin, P., Popovici, D., Larochelle, H.: greedy layer-wise training of deep networks. Advances in Neural Information Processing Systems 19, 153–160 (2007)Google Scholar
- 6.Brain Decoder Toolbox (BDTB), http://www.cns.atr.jp/dni/download/brain-decoder-toolbox/
- 7.Lancaster, J.L., Summerln, J.L., Rainey, L., Freitas, C.S., Fox, P.T.: The Talairach Daemon, a database server for Talairach Atlas Labels. NeuroImage 5, S633 (1997)Google Scholar
- 10.Schmah, T., Hinton, G., Zemel, R.S., Small, S.L., Strother, S.C.: Generative versus discriminative training of RBMs for classification of fMRI images. Advances in Neural Information Processing Systems 21, 1409–1416 (2009)Google Scholar
- 11.Lee, H., Grosse, R., Ranganath, R., Ng, A.Y.: Convolutional deep belief networks for scalable unsupervised learning of hierarchical representations. In: Proceedings of the Twenty-Sixth International Conference on Machine Learning, ICML 2009 (2009)Google Scholar