Structure to hold the parameters for the Allocation Model [5]. More...

#include <alloc_model_params.hpp>

Public Member Functions
	Params (Scalar a, Scalar b, const std::vector< Scalar > &alpha, const std::vector< Scalar > &beta)
	Construct Params manually with the given distribution parameters. More...

	Params (const shape< 3 > &tensor_shape)
	Construct Params with the default distribution parameters according to the given tensor shape. More...

	Params ()=default

Public Attributes
Scalar	a
	Shape parameter of Gamma distribution. More...

Scalar	b
	Rate parameter of Gamma distribution. More...

std::vector< Scalar >	alpha
	Parameter vector of Dirichlet prior for matrix \(W_{x \times y}\) of size \(x\). More...

std::vector< Scalar >	beta
	Parameter vector of Dirichlet prior for matrix \(H_{z \times y}\) of size \(z\). More...

Detailed Description

template<typename Scalar>
struct bnmf_algs::alloc_model::Params< Scalar >

Structure to hold the parameters for the Allocation Model [5].

According to the Allocation Model,

\[ L_j \sim \mathcal{G}(a, b) \qquad W_{:k} \sim \mathcal{D}(\alpha) \qquad H_{:j} \sim \mathcal{D}(\beta) \]

and \(S_{ijk} \sim \mathcal{PO}(W_{ik}H_{kj}L_j)\).

Template Parameters

Scalar Type of the parameters. Common choices are double and float, but integer-like types can also be used.

Constructor & Destructor Documentation

template<typename Scalar>

bnmf_algs::alloc_model::Params< Scalar >::Params	(	Scalar	a,
		Scalar	b,
		const std::vector< Scalar > &	alpha,
		const std::vector< Scalar > &	beta
	)

inline

Construct Params manually with the given distribution parameters.

Parameters

a	Shape parameter of Gamma distribution.
b	Rate parameter of Gamma distribution.
alpha	Parameter vector of Dirichlet prior for matrix \(W_{x \times y}\) of size \(x\).
beta	Parameter vector of Dirichlet prior for matrix \(H_{z \times y}\) of size \(z\).

template<typename Scalar>

bnmf_algs::alloc_model::Params< Scalar >::Params ( const shape< 3 > & tensor_shape )

inlineexplicit

Construct Params with the default distribution parameters according to the given tensor shape.

Default distribution parameters are

\begin{eqnarray*} \mbox{a} &=& 1 \\ \mbox{b} &=& 10 \\ \mbox{alpha} &=& \begin{matrix} (1 & 1 & \dots & 1) \end{matrix} \\ \mbox{beta} &=& \begin{matrix} (1 & 1 & \dots & 1) \end{matrix} \end{eqnarray*}

Parameters

tensor_shape Shape of tensor \(S_{x \times y \times z}\).

template<typename Scalar>

bnmf_algs::alloc_model::Params< Scalar >::Params ( )

default