View source on GitHub
|
Computes the covariance matrix over the whole dataset.
tft.covariance(
x: tf.Tensor, dtype: tf.DType, name: Optional[str] = None
) -> tf.Tensor
The covariance matrix M is defined as follows: Let x[:j] be a tensor of the jth element of all input vectors in x, and let u_j = mean(x[:j]). The entry M[i,j] = E[(x[:i] - u_i)(x[:j] - u_j)]. Notice that the diagonal entries correspond to variances of individual elements in the vector, i.e. M[i,i] corresponds to the variance of x[:i].
Args |
|---|
x: A rank-2 Tensor, 0th dim are rows, 1st dim are indices in each input
vector.
dtype: Tensorflow dtype of entries in the returned matrix.
name: (Optional) A name for this operation.
Raises |
|---|
ValueError: if input is not a rank-2 Tensor.
Returns |
|---|
A rank-2 (matrix) covariance Tensor