next up previous
Next: About this document ... Up: classify Previous: Suboptimal feature selection

Stories about feature selection

First story about feature selection

Once upon a time there were two farmers who lived next to each other. One day, they went to the market together and each bought a horse. As the horses ran across the fence between the two farms all the time, the farmers felt they needed to think of some way (select some features) to tell the horses apart (to classify). First they colored the ears of the two horses differently. But the color soon wore off. Then they tied a rope of different color around the neck of each horse. But then the ropes were lost. Finally, one farmer thought of a good idea and said to the other, "let's measure them. I just noticed that my red horse is quite a few inches taller than your white one!"

Second story about feature selection

The wife of a general gave birth to his son. The general sent a soldier to find out whether his son looked like him. When the soldier came back he said: Your son looks exactly like you! He is bald headed and tooth-less. He understands nothing, and he screams all the time. (The general and his newborn baby belong to the same class based on these features.)

A story related (maybe) to neural network

An elementary school teacher asked her class to give examples of the great technological inventions in the 20th century. One kid said it was the telephone, as it lets you talk to someone far away. Another kid said it was the airplane, because an airplane can take you to anywhere in the world. Then the teacher saw little Johnny eagerly waving his hand in the back of the classroom. "What do you think is the greatest invention, Johnny?" "The thermos!" The teacher was very puzzled and asked "Why do you think the thermos? All it can do is to keep warm things warm and cold things cold." "But how does it know when to keep things warm and when to keep them cold?"

Don't you sometimes wonder "How does a neural network know ...?"


next up previous
Next: About this document ... Up: classify Previous: Suboptimal feature selection
Ruye Wang 2016-11-30