拼命三娘冲（20191121)_ Neural Network v.s. Logistic Regression

Neural network learning its own features

Neural network learning is just like logistic regression, except rather than using
original features, is using new feature, which are automatically learned from the
original input. The logistic regression part in the neural network is not constrained to the raw features, instead, it feed itself with interesting
features learned from it to adapt more complex structure in data and finally can
learn more complex hypotheses.