API Reference

Tensorflow/Keras implementation of Assumed Density Filtering (ADF) based probabilistic neural networks.

This package provides implementations of several ADF based probabilistic buildings blocks commonly used in neural networks. They are to be used within the framework of Tesorflow/Keras. Unlike standard (deterministic) Keras layers that propagate point estimates, the corresponding probabilsitic ADF layers propagate a distribution parametrized by its mean and (co-)variance.

What follows is the API explanation. This mostly just lists functions and their options and is intended for quickly looking up things.

If you like a more hands-on introduction, have a look at our Examples.

kerasadf.activations

Below is a list of supported activation functions in the package.

linear(x[, mode])

Linear (identity) activation function.

relu(x[, alpha, max_value, threshold, mode])

Rectified Linear Unit.

kerasadf.layers

keras-adf layers API.

Collects aliases of all supported layers in the package. See below for details.

kerasadf.layers.convolutional

Below is a list of supported convolutional layers in the package. They have aliases making them directly available from kerasadf.layers.

Conv(rank, filters, kernel_size[, strides, ...])

Abstract nD convolution layer (private, used as implementation base).

Conv1D(filters, kernel_size[, strides, ...])

1D convolution layer (for example temporal convolution).

Conv2D(filters, kernel_size[, strides, ...])

2D convolution layer (for example spatial convolution).

kerasadf.layers.core

Below is a list of supported core layers in the package. They have aliases making them directly available from kerasadf.layers.

ADFLayer([mode])

Abstract base class for Assumed Density Filtering layers.

Dense(units[, activation, use_bias, ...])

Densly-connected (fully connected) neural network layer.

Flatten([data_format])

Flattens the input.

kerasadf.layers.pooling

Below is a list of supported pooling layers in the package. They have aliases making them directly available from kerasadf.layers.

Pooling1D(pool_function, pool_size, strides)

Pooling layer for arbitrary pooling functions, for 1D inputs.

Pooling2D(pool_function, pool_size, strides)

Pooling layer for arbitrary pooling functions, for 2D inputs.

AveragePooling1D([pool_size, strides, ...])

Average pooling layer for 1D inputs.

AveragePooling2D([pool_size, strides, ...])

Average pooling layer for 2D inputs.