How to using other models?

[masafangfang]

Recently I've knew klaus and run it on my iPhone6plus. It cost nearly 1s to handle a photo.
In your README.md you have say ccv's default image classifier follows Matt's model, then I have some personal questions:
1.Is there any paper I can read if I want to know the structure of Matt's model ?
2.If I want to use another model instead of Matt's, how should I do ? Does it possible to use the model train by Caffe or m-xnet ?
3.Klaus shows the process about how to use ccv image classifier, and can ccv be used to train a net?

[masafangfang]

1. How did you train the Matt's model and generate the image-net-2012-mobile.sqlite3 ? If I want to use another model trained by caffe, maybe should I store the parameters in the same form of this file?