today I watched vidoe of Dr. James McCaffrey about traing neural netwoks.
So, few interesting details, which was useful for me.
1. As usually it is more then enough one hidden layer.
2. Imagine, that you need to feed into your neural network some data. Let's say Age, income, sex, religion. For data normalization it is good to use not big numbers.
For example for Income which is around 51000 it's good to give to network 5.1 and not 51000.
For sex passing into NN it's good to use gues what? If you have an idea 0 and one, you are wrong. It's good to pass -1 and 1.
And for classfication like if you have input for example 3 different categories, the best way to pass not values like 1, 2, 3, but to pass three inputs, and pass at them 100, 010 or 001. For example if you need to code Christian, Muslim, Buddhist, you'll pass 100 for Christian, 010 for Muslim, 001 for Buddhist.
3. Four most common activation functions:
a. logistic sigmoid
b. hyperbolic tangent
c. Heaviside step
4. Training algorithms for neural networks:
b. genetic algorithm
c. particle swarm optimization