Given an arbitrary set of symbols (the english alphabet is the example that will be used here), Huffman coding is a way of creating the most efficient (smallest) binary code for that set of symbols. It exploits the fact that in most data some symbols