Network parameters to estimate average of a unidimensional variable

Hello Yamada,

I found your implementation of the Probabilistic Backpropagation algorithm super helpful. I wish to know your opinion on the following problem- Suppose I have an exponential data generating process whose true mean is 10.

import numpy as np
from scipy import stats
avg = 10
n = 100
X = np.ones(n).reshape(-1,1)
y= stats.expon(avg).rvs(n)

How can I use your code to get the mean and variance of the estimated mean parameter?

My current implementation is
_X = X
_y = y.reshape(-1,1)
pbp = b4tf.models.PBP([1],input_shape=X.shape[1])
pbp.fit(_X,_y,batch_size=8)

Would be glad if you can share your opinion on this.