Since I'm implementing a project in "Face Detection", I have to use advanced methods for "Feature Extraction". I have designed a MLP network for training, but I used raw pictures as an input layer… As maybe you know, it's better idea to decrease the input dimension by using Feature Extraction/Selection approaches.
There is some approaches like :"non Linear PCA" and "Kernel Methods".
I want to use Kernel Methods. I think it's more applicable and most of recent papers that I have read were about this major! However , It's maybe because of my guide attitude (Dr.Araabi) who have focused on it more :) I don't know. But because my learning procedure is "Learning by Doing" I will implement both of them. If you have any prior experience, please email me and let me know…
Cheers :)
