Artificial Intelligence By Example

Covariance

Our goal in this section is to find out whether two words, for example, will often be found together or close to each other, taking the output of the embedding program into account.

Covariance will tell us whether these datasets vary together or not. The equation follows the same philosophy as variance, but now both variances are joined to see whether they belong together or not:

As with the variance, the denominator can be n-1 depending on your model. Also, in this equation, the numerator is expanded to visualize the co-part of covariance, as implemented in the following array in math.py:

x=np.array([[1, 2, 3, 4], [1, 2, 3, 5]]) ...