I am curious to if there is data available for MLP architectures in use today, their initial architecture, the steps that were taken to improve the architecture to an acceptable state and what the problem is the neural network aimed to solve.
For example, what the initial architecture (amount of hidden layers, the amount of neurons) was for a MLP in a CNN, the steps taken to optimize the architecture (adding more layers and reducing nodes, changing activation functions) and the results each step produced (i.e. increased error or decreased error). What the problem is the CNN tried to solve (differentiation of human faces, object detection inteded for self driving cars etc.)
Of course I used a CNN as an example but I am referring to data for any MLP architecture in plain MLPs or Deep Learning architectures such as RNNs, CNNs and mroe. I am focused on the MLP architecture mostly.
If there is not how do you think one can accumulate this data?
