1. Demand Prediction

• The features are combined to represent a factor that determines the output, which are then combined to output a single value.
• These factors are known as activations.
  
• Could be said that the vectors become a scalar value in the end.
• Each layer has single or multiple neurons.
• The layer with features are known as input layer.
• The layers that compute values in the middle are hidden layers.
• The layer with the final output is known as output layer.
  
• There could be multiple hidden layers.
• The neural network with multiple layers is known as multilayer perceptron.

2. Example: Face Recognition

• The picture's pixels with intensity values are spread out in a vector for input layer.
  
• The neural network starts from recognizing smaller regions of the pic to bigger.
• These feature detectors for hidden layers are learnt all by itself.
• Activations are "higher level features".
  
• Recognizing a car also takes similar steps.
• Starting with smaller parts of the car to bigger.