# Bayesian multilayer perceptron

**URL:** <https://discourse.pymc.io/t/bayesian-multilayer-perceptron/4362>\
**Category:** Questions\
**Created:** [January 15, 2020, 9:35am UTC](https://discourse.pymc.io/t/bayesian-multilayer-perceptron/4362 "2020-01-15T09:35:43Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![SergeyMalashenko](https://avatars.discourse-cdn.com/v4/letter/s/a4c791/32.png) [@SergeyMalashenko](https://discourse.pymc.io/u/SergeyMalashenko)\
**Post date:** [January 15, 2020, 9:35am UTC](https://discourse.pymc.io/t/bayesian-multilayer-perceptron/4362/1 "2020-01-15T09:35:43Z")

</div>

Good day! I implemented Bayesian neural network. All weights have normal distribution. I know that input data has different distributions for every feature ( normal, multinomial and soon). So my question is it a right idea to have conjunction distributions in the first layer.