Bloom Filter

2. Dev Learning

Bloom Filter

From Big Data notes, Bloom Filter:

• Keep an array $A$ of $m$ bits
• With $k$ hash functions, do $A[h_j(e)]=1,\forall 1\le j\le k$
• For the membership test, $x_i\in S$ if $A[h_1(x_1)]=...=A[h_k(x_i)]=1$
• probability that $x_i$ is erroneously claimed to be in $S$ is: $P(A[h_j(x_i)]=1,\forall 1\le j\le k)\simeq (1-e^{-km/n}{)}^k$

-> typically $k$ is a small constant which depends on the desired false error rate $ϵ$
-> $m$ is proportional to $k$

Insertion:

• feed the element to the $k$ hash functions to get $k$ array positions
• set positions bits to 1
  Testing:
• feed the element to the $k$ hash functions to get $k$ array positions
• two cases
  • any of the position is at 0 = element def not in the set
  • otherwise = either:
    • the element is in the set
    • bits have by chance been set to 1 during insertion of other elements (no way to tell the difference)

—> This is the simple Bloom Filter.