MindSpore Probability Python API

Probability.

The high-level components used to construct the probabilistic network.

Bayesian Layers

API Name	Description	Supported Platforms
mindspore.nn.probability.bnn_layers.ConvReparam	Convolutional variational layers with Reparameterization.	Ascend GPU
mindspore.nn.probability.bnn_layers.DenseLocalReparam	Dense variational layers with Local Reparameterization.	Ascend GPU
mindspore.nn.probability.bnn_layers.DenseReparam	Dense variational layers with Reparameterization.	Ascend GPU

Prior and Posterior Distributions

API Name	Description	Supported Platforms
mindspore.nn.probability.bnn_layers.NormalPosterior	Build Normal distributions with trainable parameters.	Ascend GPU
mindspore.nn.probability.bnn_layers.NormalPrior	To initialize a normal distribution of mean 0 and standard deviation 0.1.	Ascend GPU

Bayesian Wrapper Functions

API Name	Description	Supported Platforms
mindspore.nn.probability.bnn_layers.WithBNNLossCell	Generate a suitable WithLossCell for BNN to wrap the bayesian network with loss function.	Ascend GPU

Deep Probability Networks

API Name	Description	Supported Platforms
mindspore.nn.probability.dpn.ConditionalVAE	Conditional Variational Auto-Encoder (CVAE).	Ascend GPU
mindspore.nn.probability.dpn.VAE	Variational Auto-Encoder (VAE).	Ascend GPU

Infer

API Name	Description	Supported Platforms
mindspore.nn.probability.infer.ELBO	The Evidence Lower Bound (ELBO).	Ascend GPU
mindspore.nn.probability.infer.SVI	Stochastic Variational Inference(SVI).	Ascend GPU

ToolBox

API Name	Description	Supported Platforms
mindspore.nn.probability.toolbox.UncertaintyEvaluation	Toolbox for Uncertainty Evaluation.	Ascend GPU
mindspore.nn.probability.toolbox.VAEAnomalyDetection	Toolbox for anomaly detection by using VAE.	Ascend GPU

Model Transformer

API Name	Description	Supported Platforms
mindspore.nn.probability.transforms.TransformToBNN	Transform Deep Neural Network (DNN) model to Bayesian Neural Network (BNN) model.	Ascend GPU