for my master thesis I want to predict three variables (Arousal, Valence, Dominance) from so called salience maps. I am running a multiple output CNN on the salience maps. However, the model is pretty much overfitting, regardless of the tuning of learning rate, batch size or complexity.
I am actually not sure if I did one of the more simple things wrong - the standarization.
The descriptive statistics for my X look like this:
Mean = 0.05043895889652882,
SD = 0.055531537078160255,
Do I still need to standardize the variable? I am a bit confused here, because I am kind of new to Deep Learning and am getting lost in the jungle of solutions for my overfitting model.
Please let me know if I can provide additional information.