12/1/17

I was getting a bit tired of the data analysis sections of my first udemy course and felt like it might be good to mix it up a bit, so I watched the first 6 lectures on deep learning. The first lecture was on the neuron; the lecture connected the biology of the neuron to the use of artificial neurons in a neural network to simulate the structure/operation of the human brain. The lecture also showed how a signal propagates through an artificial neuron through the use of weighted sums. The second lecture discussed activation functions, which are used to determine whether or not a neuron "fires" when given input, just as a neuron in the human brain fires only when the electrical signal from the previous neuron is strong enough. The third lecture was a great walkthrough of an example involving using a neural network to perform a house price valuation, which really showed the power of neural networks and reinforced my intuition. The fourth lecture was even better, showing how a basic neural network (3 inputs: study hours, quiz grade, and sleep hours) could be fed into a neural network and produce an output which would be evaluated on a cost function, which tells the network how close the predicted value it produced was to the actual value in the data.