what kind of neural network is suitable for authentication

Sir,

I need to know about which type of neural network is suitable for authentication. I have the inputs of username, password and clickpoints of the image. If I pass both username and password also clickpoints of the image, then only the user authentication is valid otherwise invalid. For this information, I need the above inputs to train neural network. I don't know about which is network is suitable for my inputs.

Answer 1

[Krishna]

Hi,

Seems like you want to build a face recognition system along with password control. Validating password with usernames is not a deep learning problem for that you can check out,

https://in.mathworks.com/matlabcentral/answers/402747-creating-a-password-checker-using-a-function

For face recognition systems you need to build a similarity function using pre-trained architectures like ‘googleNet’ or ‘ResNet.’ This similarity function checks whether two images are same or not by comparing the final outputs of fully connected layer in the backbone architecture that you have used. To learn more, you can search for Siamese network or One-shot learning networks and learn how they work.

Also, you can look at these similar examples provided by MathWorks,

https://in.mathworks.com/help/deeplearning/ug/train-twin-network-to-compare-images.html

Hope this helps.