ML::SVM::linear Namespace Reference

Classes
struct	Params

Functions
template<typename T >
int	fit (const raft::handle_t &handle, const Params &params, const std::size_t nRows, const std::size_t nCols, const int nClasses, const T *classes, const T *X, const T *y, const T *sampleWeight, T *w, T *probScale)
	Fit a linear SVM model. More...
template<typename T >
void	computeProbabilities (const raft::handle_t &handle, const std::size_t nRows, const int nClasses, const T *probScale, T *scores, T *out)
	Compute probabilities from decision function scores. More...

Function Documentation

computeProbabilities()

template<typename T >

void ML::SVM::linear::computeProbabilities	(	const raft::handle_t &	handle,
		const std::size_t	nRows,
		const int	nClasses,
		const T *	probScale,
		T *	scores,
		T *	out
	)

Compute probabilities from decision function scores.

Parameters

[in]	handle	the cuML handle.
[in]	nRows	the number of input samples.
[in]	nClasses	the number of input classes.
[in]	probScale	the probability scales, shape=(nClasses, 2), F-contiguous.
[in,out]	scores	the decision function scores, shape=(nRows, nClasses), C-contiguous. Note that this array will be mutated in-place during the calculation.
[out]	out	the computed probabilities, shape=(nRows, nClasses), C-contiguous.

fit()

template<typename T >

int ML::SVM::linear::fit	(	const raft::handle_t &	handle,
		const Params &	params,
		const std::size_t	nRows,
		const std::size_t	nCols,
		const int	nClasses,
		const T *	classes,
		const T *	X,
		const T *	y,
		const T *	sampleWeight,
		T *	w,
		T *	probScale
	)

Fit a linear SVM model.

Parameters

[in]	handle	the cuML handle.
[in]	params	the model parameters.
[in]	nRows	the number of input samples.
[in]	nCols	the number of feature dimensions.
[in]	nClasses	the number of input classes, or 0 for a regression problem.
[in]	classes	the unique input classes, shape=(nClasses,), or nullptr for a regression problem.
[in]	X	the training data, shape=(nRows, nCols), F-contiguous
[in]	y	the target data, shape=(nRows,)
[in]	sampleWeight	non-negative weights for the training data, shape=(nRows,), or nullptr if unweighted.
[out]	w	the fitted weights, shape=(nCoefs, nCols) or (nCoefs + 1, nCols + 1) if fit_intercept=true, where nCoefs = 1 for regression or if nClasses = 2, and nClasses otherwise. F-contiguous.
[out]	probScale	the fitted probability scales, shape=(nClasses, 2), F-contiguous. Pass nullptr to not fit probability scales.

Returns

n_iter: the maximum number of iterations run across all classes.