They 8216;consider8217; that activation maps obtained in the intermediate (hidden) layers will follow Gaussian Mixture Model.
Notation: They say the target mixture model will contain S Gaussians.
So far it’s okay for me.
Then, in Section IV. Functional Watermarking, they say, in practice they select S to be number of classes in the classification task.
This is where I get confused. If you can have one Gaussian for each class already at an intermediate layer, why would you need any additional layers? (Alternatively, if you cannot have a Gaussian for each class at the given layer, then such selection looks too restrictive.)
Do I miss something?
thanks you RSS link
More link Blog tech
more link ADS
Blockchain, bitcoin, ethereum, blockchain technology, cryptocurrencies
Information Security, latest Hacking News, Cyber Security, Network Sec
Information Security, latest Hacking News, Cyber Security, Network Security
Blog! Development Software and Application Mobile
Development apps, Android, Ios anh Tranning IT, data center, hacking
Car News, Reviews, Pricing for New & Used Cars, car reviews and news, concept cars