August 2018
Intermediate to advanced
522 pages
12h 45m
English
If X is a Bernoulli-distributed random variable, it can have only two possible outcomes (for simplicity, let's call them 0 and 1) and their probability is this:

In general, the input vectors xi are assumed to be multivariate Bernoulli distributed and each feature is binary and independent. The parameters of the model are learned according to a frequency count. Hence, if there are n samples with m features, the probability for the ith feature is this (Nx(i) counts the number of times the ith = 1):

To test this algorithm ...
Read now
Unlock full access