Training SVM with features vectors of different sizes
2 ビュー (過去 30 日間)
I want to train SVM to classify a normal and abnormal groups. To do that I supposed to extract the features using LoG with invariant scale. The problem is that the resulting features are vary in size. I added all of these features into a cell array. But SVM in MATLAB do not accept cell arrays as an input for the predictor data parameter (x) in the fitcsvm function Mdl = fitcsvm(X,Y). How can I solve this problem, please?
I was thinking of defining a zero matrix of rows (the samples) and columns (max feature size). Then concatenate the features in this matrix. But I do not know if this will affect the SVM training procedure. Any suggestions, please?
Thank you in advance