What layers to use for data classifcation?

Hi,

I am very new to MATLAB. I am using it to (hopefully) create a neural network to classify images with or without steganography. I have created histograms for each of these images, half or which are stegged, and half not. I have then put that data into a 256x3 matrix for each of the images, the 3 being each colour channel of the image. I want to be able to output a value of between 1 and 0 as that will be able to determine whether an image is stegged or not. The features I want to pick up on is that on the stegged images the histogram data fluctuates along the graph line, where the normal images don't. I understand I'll need a sequence input layer, however from there I am completely lost and have no idea what layer(s) to use next.

Many thanks

Answer 1

[Srivardhan Gadila]

Refer to Deep Learning Toolbox — Examples and Deep Learning Toolbox to get started.

You can use an imageInputLayer or an sequenceInputLayer followed by Convolution and Fully Connected Layers or Sequence Layers with other layers like Activation, Normalization, Pooling layers etc. followed by softmaxLayer and classificationLayer at the end.

Refer to List of Deep Learning Layers.