Understanding the distribution shapes

[Gurkirat Singh]

Hi there, I want confirm my understanding of the event, batch and sample shapes that I have learnt from here

Here are the points

- sampling is selection of events from a population / distribution

- identically distributed samples means, each element is from same distribution (same properties like mean and std)

- event shape is the shape of the single event from single sampling of a single distribution

- batch shape is the shape of the single event from single sampling of different distribtions

- sample shape is the shape of the events from single sampling of the distributions

so if there are two distributions with each event shape (2, 3), and you sample 2 events (experiment is done now its time of output), then the shape would be (2, 2, 2, 3)

2 2 2 3

^ ^ ^ ^

| | ---

|_|__|____________ sample shape (selection of subsets from pool of sample space of experiment)

|__|____________ batch shape (independent and unidentical distributions, defined by experiments)

|____________ event shape (single event in the sample space, where |E| = 1 and E ⊂ S)

[Gurkirat Singh]

Here is the image version of what I want to say

[rif]

This seems essentially correct to me, although I didn't 100% understand every single sentence.

--
You received this message because you are subscribed to the Google Groups "TensorFlow Probability" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tfprobabilit...@tensorflow.org.
To view this discussion on the web visit https://groups.google.com/a/tensorflow.org/d/msgid/tfprobability/5eac7b69-0a0a-47f2-a683-c4ddac4b68c1n%40tensorflow.org.