Neural network classifier for classification of brain MRI Images

1 次查看（过去 30 天）
purti choudhary
purti choudhary 2015-5-20
回答： Greg Heath 2015-5-20
I have a data set of total 35 images, 20 used for training and rest 15 for testing. Two class labels normal and abnormal. Total ten texture features namely energy, entropy, contrast, cluster shade, covariance, inertia, cluster prominance, cluster shade, idm, angular second moment are extracted. How to use neural network for this classification purpose.

请先登录，再回答此问题。

采纳的回答

Greg Heath
Greg Heath 2015-5-20
It doesn't sound like you have a sufficient amount of data for precision estimation.
Therefore I would recommend many repetitions of crossvalidation using patternnet. Combining validation stopping and regularization might be worthwhile.
[ I N ] = size(input) % [ 10 35 ]
[ O N ] = size(target) % [ 1 35 ]
With no data division, the number of unknown weights will exceed the number of training equations when the number of hidden nodes exceeds the upper bound
Hub = -1 + ceil( (N*O-O)/ (I+O+1) ) = 2
Therefore you may have to experiment more than you would like to with input variable reduction and regularization.
Hope this helps.
Greg

更多回答（0 个）

类别

在 Help Center 和 File Exchange 中查找有关 Deep Learning Toolbox 的更多信息

标签

产品

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by