ML::GLM::opg Namespace Reference

Functions
void	preProcessData (raft::handle_t &handle, std::vector< MLCommon::Matrix::Data< float > * > &input_data, MLCommon::Matrix::PartDescriptor &input_desc, std::vector< MLCommon::Matrix::Data< float > * > &labels, float *mu_input, float *mu_labels, float *norm2_input, bool fit_intercept, bool normalize, cudaStream_t *streams, int n_streams, bool verbose)
void	preProcessData (raft::handle_t &handle, std::vector< MLCommon::Matrix::Data< double > * > &input_data, MLCommon::Matrix::PartDescriptor &input_desc, std::vector< MLCommon::Matrix::Data< double > * > &labels, double *mu_input, double *mu_labels, double *norm2_input, bool fit_intercept, bool normalize, cudaStream_t *streams, int n_streams, bool verbose)
void	postProcessData (raft::handle_t &handle, std::vector< MLCommon::Matrix::Data< float > * > &input_data, MLCommon::Matrix::PartDescriptor &input_desc, std::vector< MLCommon::Matrix::Data< float > * > &labels, float *coef, float *intercept, float *mu_input, float *mu_labels, float *norm2_input, bool fit_intercept, bool normalize, cudaStream_t *streams, int n_streams, bool verbose)
void	postProcessData (raft::handle_t &handle, std::vector< MLCommon::Matrix::Data< double > * > &input_data, MLCommon::Matrix::PartDescriptor &input_desc, std::vector< MLCommon::Matrix::Data< double > * > &labels, double *coef, double *intercept, double *mu_input, double *mu_labels, double *norm2_input, bool fit_intercept, bool normalize, cudaStream_t *streams, int n_streams, bool verbose)
template<typename T >
std::vector< T >	getUniquelabelsMG (const raft::handle_t &handle, Matrix::PartDescriptor &input_desc, std::vector< Matrix::Data< T > * > &labels)
	Calculate unique class labels across multiple GPUs in a multi-node environment. More...
template<typename T >
void	qnFit (raft::handle_t &handle, std::vector< Matrix::Data< T > * > &input_data, Matrix::PartDescriptor &input_desc, std::vector< Matrix::Data< T > * > &labels, T *coef, const qn_params &pams, bool X_col_major, bool standardization, int n_classes, T *f, int *num_iters)
	performs MNMG fit operation for the logistic regression using quasi newton methods More...
template<typename T , typename I >
void	qnFitSparse (raft::handle_t &handle, std::vector< Matrix::Data< T > * > &input_values, I *input_cols, I *input_row_ids, I X_nnz, Matrix::PartDescriptor &input_desc, std::vector< Matrix::Data< T > * > &labels, T *coef, const qn_params &pams, bool standardization, int n_classes, T *f, int *num_iters)
	support sparse vectors (Compressed Sparse Row format) for MNMG logistic regression fit using quasi newton methods More...

Function Documentation

getUniquelabelsMG()

template<typename T >

std::vector<T> ML::GLM::opg::getUniquelabelsMG	(	const raft::handle_t &	handle,
		Matrix::PartDescriptor &	input_desc,
		std::vector< Matrix::Data< T > * > &	labels
	)

Calculate unique class labels across multiple GPUs in a multi-node environment.

Parameters

[in]	handle	the internal cuml handle object
[in]	input_desc	PartDescriptor object for the input
[in]	labels	labels data

Returns

host vector that stores the distinct labels

postProcessData() [1/2]

void ML::GLM::opg::postProcessData	(	raft::handle_t &	handle,
		std::vector< MLCommon::Matrix::Data< double > * > &	input_data,
		MLCommon::Matrix::PartDescriptor &	input_desc,
		std::vector< MLCommon::Matrix::Data< double > * > &	labels,
		double *	coef,
		double *	intercept,
		double *	mu_input,
		double *	mu_labels,
		double *	norm2_input,
		bool	fit_intercept,
		bool	normalize,
		cudaStream_t *	streams,
		int	n_streams,
		bool	verbose
	)

postProcessData() [2/2]

void ML::GLM::opg::postProcessData	(	raft::handle_t &	handle,
		std::vector< MLCommon::Matrix::Data< float > * > &	input_data,
		MLCommon::Matrix::PartDescriptor &	input_desc,
		std::vector< MLCommon::Matrix::Data< float > * > &	labels,
		float *	coef,
		float *	intercept,
		float *	mu_input,
		float *	mu_labels,
		float *	norm2_input,
		bool	fit_intercept,
		bool	normalize,
		cudaStream_t *	streams,
		int	n_streams,
		bool	verbose
	)

preProcessData() [1/2]

void ML::GLM::opg::preProcessData	(	raft::handle_t &	handle,
		std::vector< MLCommon::Matrix::Data< double > * > &	input_data,
		MLCommon::Matrix::PartDescriptor &	input_desc,
		std::vector< MLCommon::Matrix::Data< double > * > &	labels,
		double *	mu_input,
		double *	mu_labels,
		double *	norm2_input,
		bool	fit_intercept,
		bool	normalize,
		cudaStream_t *	streams,
		int	n_streams,
		bool	verbose
	)

preProcessData() [2/2]

void ML::GLM::opg::preProcessData	(	raft::handle_t &	handle,
		std::vector< MLCommon::Matrix::Data< float > * > &	input_data,
		MLCommon::Matrix::PartDescriptor &	input_desc,
		std::vector< MLCommon::Matrix::Data< float > * > &	labels,
		float *	mu_input,
		float *	mu_labels,
		float *	norm2_input,
		bool	fit_intercept,
		bool	normalize,
		cudaStream_t *	streams,
		int	n_streams,
		bool	verbose
	)

qnFit()

template<typename T >

void ML::GLM::opg::qnFit	(	raft::handle_t &	handle,
		std::vector< Matrix::Data< T > * > &	input_data,
		Matrix::PartDescriptor &	input_desc,
		std::vector< Matrix::Data< T > * > &	labels,
		T *	coef,
		const qn_params &	pams,
		bool	X_col_major,
		bool	standardization,
		int	n_classes,
		T *	f,
		int *	num_iters
	)

performs MNMG fit operation for the logistic regression using quasi newton methods

Parameters

[in]	handle	the internal cuml handle object
[in]	input_data	vector holding all partitions for that rank
[in]	input_desc	PartDescriptor object for the input
[in]	labels	labels data
[out]	coef	learned coefficients
[in]	pams	model parameters
[in]	X_col_major	true if X is stored column-major
[in]	standardization	whether to standardize the dataset before training
[in]	n_classes	number of outputs (number of classes or 1 for regression)
[out]	f	host pointer holding the final objective value
[out]	num_iters	host pointer holding the actual number of iterations taken

qnFitSparse()

template<typename T , typename I >

void ML::GLM::opg::qnFitSparse	(	raft::handle_t &	handle,
		std::vector< Matrix::Data< T > * > &	input_values,
		I *	input_cols,
		I *	input_row_ids,
		I	X_nnz,
		Matrix::PartDescriptor &	input_desc,
		std::vector< Matrix::Data< T > * > &	labels,
		T *	coef,
		const qn_params &	pams,
		bool	standardization,
		int	n_classes,
		T *	f,
		int *	num_iters
	)

support sparse vectors (Compressed Sparse Row format) for MNMG logistic regression fit using quasi newton methods

Parameters

[in]	handle	the internal cuml handle object
[in]	input_values	vector holding non-zero values of all partitions for that rank
[in]	input_cols	vector holding column indices of non-zero values of all partitions for that rank
[in]	input_row_ids	vector holding row pointers of non-zero values of all partitions for that rank
[in]	X_nnz	the number of non-zero values of that rank
[in]	standardization	whether to standardize the dataset before training
[in]	input_desc	PartDescriptor object for the input
[in]	labels	labels data
[out]	coef	learned coefficients
[in]	pams	model parameters
[in]	n_classes	number of outputs (number of classes or 1 for regression)
[out]	f	host pointer holding the final objective value
[out]	num_iters	host pointer holding the actual number of iterations taken