summary_Deep_Learning.txt

Deep Learning uses a Neural Network to imitate animal intelligence.
There are three types of layers of neurons in a neural network: the Input Layer, the Hidden Layer(s), and the Output Layer.
Connections between neurons are associated with a weight, dictating the importance of the input value.
Neurons apply an Activation Function on the data to “standardize” the output coming out of the neuron.
To train a Neural Network, you need a large data set.
Iterating through the data set and comparing the outputs will produce a Cost Function, indicating how much the AI is off from the real outputs.
After every iteration through the data set, the weights between neurons are adjusted using Gradient Descent to reduce the cost function.