Is it possible to download the tables of feature vectors that Core ML or Create Create ML extracts after applying an image classifier to extract features from images? I understand the models that are being used (SqueezeNet, ResNet, ScreenPrint, etc), but I'd like to be able to download the features that were extracted, to use in other platforms.
I want to perform scientific experiments using these Core ML and Create ML, but if I cannot review the extracted features, therefore, I cannot validate the work externally. I believe both platforms are very powerful for applied research purposes, but the methods and APIs are not open source, so the "black box" nature of the process is hard to describe scientifically.
Any guidance how I can full cite, and obtain the detail I am looking for would be greatly appreciated, thanks.