Entropy
Last updated
Was this helpful?
Last updated
Was this helpful?
Let's say I want to send the result of 10 coin-flips with only 0 and 1(which is called bit). We can send it by sending 1 if Head appears, and 1 if Tail appears.
Another example is sending an number between 1~100.
Some might answer that we have to send 100 bits. Sending an answer of following 100 questions: is it 1, is it 2, is it 3, ... is it 100.
Clever people would answer . This could be achieved by sending answers of following questions: is it bigger than 50? is it bigger than 25/75? ...
Entropy is the quantity of the information. In other words, # of required bits to represent the information.
In the examples at Information theory, I only mentioned about the case where possiblities of every cases are even. Coin can have Head or Tail by 50%, 50% probability.
How can we express if the Head / Tail probability is 30% / 70%?
Would it be ? No. Since we know that Tail occur more often, simply representing it as is it Tail?
isn't optimal to compact the informations into bits.
Let's look at the example. Machine A, and Machine B
Machine A outputs character "A", "B", "C", "D" in equal probability, 0.25.
Machine B outputs character "A", "B", "C", "D" in 0.5, 0.125, 0.125, 0.25 probability.
Since the probability is different, we can't do the same questions as machine A.
Instead, we can question as follows:
In the probability distribtution, entropy can be expressed as follows:
If the probability distribution is even, then we need more bits(a.k.a # of questions) to express the data. However, if the probability distribution is not even, then lesser bits are required to express the data.
Amount of bits required can be calculated as
Amout of bits required can be calculated as
[1]
[2]
[3]