What is the advantage to using bloom filters?

Question

I am reading up on bloom filters and they just seem silly. Anything you can accomplish with a bloom filter, you could accomplish in less space, more efficiently, using a si

Answer 1

Bloom filters are quite useful in bioinformatics. They can be more space efficient compared to using a regular hash, especially when the size of the strings you are working with can be hundreds of millions of letters with a very small alphabet ie {A,G,T,C} . They are usually used to assess whether a certain k-mer is present or absence in a genome. There's an example of one used for something relevant here.

EDIT:

The multiple hash functions are used to minimize false positives. The hope is that between all of the k-hash functions each value will have a unique signature in the bit-array compared to every other possible value. However, false positives do exist, but they can be minimized to a manageable level. Using this technique you hash elements independently of their size. When you search for them, you use each hash function and check to make sure their bit-values are all 1.

Compare this to the human genome, where an increase in the size of the element increases the size of the hash table significantly (The table size is 4*4^k). This is assuming you encode the the elements using 2 bits / letter.