I have a small dataset with about 210 data points for training with 5 input and 2 target parameters.
The dataset is about tile production, the input parameters are the production process parameters of the tiles and the output is the physical strength and water resistance of the tiles.
I am training two multilayer perceptrons for the two target parameters, but when I try to optimize the hyperparameters of the model with a random search, the optimal parameters found are the same for both models. I am using Scikit-learn and I am trying to optimize the number of hidden layers and nodes in the model, the solver algorithm used, the initial learning rate, and the regularization term.