CODEWITHSUNDEEP
tensorflowtensorflow-probability
Anil Gaddam
Anil Gaddam

Reputation: 9

Multiply distributions in tensorflow

I am new to Tensorflow and I want to multiply two distributions to get posterior density. How can I do it using tensorflow?

For example:

likelihood = tf.contrib.distributions.MultivariateNormalDiag(loc = [0., 0., 0.], scale_diag= [1., 1., 1.])
prior = tf.contrib.distributions.MultivariateNormalDiag(loc = [0., 0., 0.], scale_diag= [1., 1., 1.])

I tried using tf.multiply(likelihood,prior) but it gives me datatype error

Failed to convert object of type to Tensor. Contents: tf.distributions.MultivariateNormalDiag("MultivariateNormalDiag", batch_shape=(), event_shape=(3,), dtype=float32). Consider casting elements to a supported type.

Can anyone please help me with this.

Help much appreciated. Thanks

Upvotes: 0

Views: 413

Answers (1)

srvasude
srvasude

Reputation: 161

A tf.distribution is an object, and hence can not be used as a Tensor.

You can instead multiply (or sum) the .prob (.log_prob) methods.

Upvotes: 1

Related Questions