Supervised Learning - Classification Week 9 Challenge

1. Neural Network is a subset of Machine learning and main logic engine for deep learning which mimics the activity of human brain.It is also called Artificial Neural Network.
Artificial Neural Network contains a node layer, an input layer, one or more hidden layers and an output layer. Node is similar to neurons in human brain also called artificial neuron. Nodes are connected to each other with corresponding weight and threshold. If a node is above the specified threshold, the corresponding node would be activated and sends data to next layer.

2. Deep learning is a subset of Machine learning inspired my structure and function of human brain. Deep learning improves automation, analytical and autonomous behaviour of a system which substantially contributes to human society, But achieve these incredible outputs, Deep learning has a trade off like it requires large number of labeled data and huge computational power.

3. Activation function plays an important role in building the artificial neural network, which has great impact in selecting hidden layer and output layer of the network. Neurons in the artificial neural network works in correspondence with weight, bias and their respective activation function. Back-propogation in the artificial neural network is possible due to activation function.

4. Linear function, Sigmoid function, Tanh function, RELU function, Softmax function.

5. Based on error at the output, weights and biasis of neuron in the hidden layer would be updated this proess is called Back propagation. Activation function supplies gradient along with error to update weight and biasis to perfom back propogation. Updating weight and biasis is important for building good Artificial Neural Network.