Op_kmeans

Data Structures
class	ecocpak::op_kmeans
Functions
template<typename eT >
static eT	ecocpak::op_kmeans::compute_distance (const subview_row< eT > &sample_in, const subview_row< eT > &centroid_in)
template<typename eT >
static void	ecocpak::op_kmeans::compute_distance (Row< eT > &dist_out, u32 &winner, const Mat< eT > &centroids_in, const subview_row< eT > &sample_in)
template<typename eT >
static void	ecocpak::op_kmeans::init_centroids (Mat< eT > &centroids_out, const Mat< eT > &samples_in, const u32 k)
	Initialize centroids.
template<typename eT >
static void	ecocpak::op_kmeans::direct_kmeans (Col< u32 > &indices_out, Col< u32 > &ranks, Mat< eT > &centroids_out, const Mat< eT > &in, const u32 k, const eT lr)
	K-means clustering.
template<typename eT >
static eT	op_kmeans::compute_distance (const subview_row< eT > &sample_in, const subview_row< eT > &centroid_in)
template<typename eT >
static void	op_kmeans::compute_distance (Row< eT > &dist_out, u32 &winner, const Mat< eT > &centroids_in, const subview_row< eT > &sample_in)
template<typename eT >
static void	op_kmeans::init_centroids (Mat< eT > &centroids_out, const Mat< eT > &samples_in, const u32 k)
	Initialize centroids.
template<typename eT >
static void	op_kmeans::direct_kmeans (Col< u32 > &indices_out, Col< u32 > &ranks, Mat< eT > &centroids_out, const Mat< eT > &in, const u32 k, const eT lr)
	K-means clustering.

---

Function Documentation

template<typename eT >

eT op_kmeans::compute_distance	(	const subview_row< eT > &	sample_in,
		const subview_row< eT > &	centroid_in
	)		[inline, static]

• Calculate distance between one sample and one centroid (returns a scalar).

• Input Arguments:
  • sample_in : Subview row vector which represents one sample.
  • centroid_in : Subview row vector which represents the centroid.

• Output Argument:
  • Scalar which denotes the squared euclidean distance between sample_in and centroid_in.

template<typename eT >

void op_kmeans::compute_distance	(	Row< eT > &	dist_out,
		u32 &	winner,
		const Mat< eT > &	centroids_in,
		const subview_row< eT > &	sample_in
	)		[inline, static]

• Compute distances between one sample and all the centroids.

• Input Arguments:
  • cetroids_in : Matrix of the centroids (i.e., each row represents a centroid).
  • sample_in : Row vector which represents the sample.

• Output Arguments :
  • dist_out : Row vector of input sample's distances from each one of the centroids.
  • winner : Index which indicates the winning centroid.

template<typename eT >

eT op_kmeans::compute_distance	(	const subview_row< eT > &	sample_in,
		const subview_row< eT > &	centroid_in
	)		[inline, static]

• Calculate distance between one sample and one centroid (returns a scalar).

• Input Arguments:
  • sample_in : Subview row vector which represents one sample.
  • centroid_in : Subview row vector which represents the centroid.

• Output Argument:
  • Scalar which denotes the squared euclidean distance between sample_in and centroid_in.

template<typename eT >

void op_kmeans::compute_distance	(	Row< eT > &	dist_out,
		u32 &	winner,
		const Mat< eT > &	centroids_in,
		const subview_row< eT > &	sample_in
	)		[inline, static]

• Compute distances between one sample and all the centroids.

• Input Arguments:
  • cetroids_in : Matrix of the centroids (i.e., each row represents a centroid).
  • sample_in : Row vector which represents the sample.

• Output Arguments :
  • dist_out : Row vector of input sample's distances from each one of the centroids.
  • winner : Index which indicates the winning centroid.

template<typename eT >

void op_kmeans::direct_kmeans	(	Col< u32 > &	indices_out,
		Col< u32 > &	ranks,
		Mat< eT > &	centroids_out,
		const Mat< eT > &	samples_in,
		const u32	k,
		const eT	lr
	)		[inline, static]

K-means clustering.

• Input Arguments:
  • samples_in : Samples' matrix (i.e., each row is a sample).
  • k : k denotes the number of desired clusters.
  • lr : Starting learning rate for online phase version of K-means.

• Output Arguments:
  • indices_out : Cluster indices for each point.
  • ranks_out : Vector which holds the number of samples that correspond to each centroid.
  • centroids_out : Cluster centroid locations.

template<typename eT >

void op_kmeans::direct_kmeans	(	Col< u32 > &	indices_out,
		Col< u32 > &	ranks,
		Mat< eT > &	centroids_out,
		const Mat< eT > &	samples_in,
		const u32	k,
		const eT	lr
	)		[inline, static]

K-means clustering.

• Input Arguments:
  • samples_in : Samples' matrix (i.e., each row is a sample).
  • k : k denotes the number of desired clusters.
  • lr : Starting learning rate for online phase version of K-means.

• Output Arguments:
  • indices_out : Cluster indices for each point.
  • ranks_out : Vector which holds the number of samples that correspond to each centroid.
  • centroids_out : Cluster centroid locations.

template<typename eT >

void op_kmeans::init_centroids	(	Mat< eT > &	centroids_out,
		const Mat< eT > &	samples_in,
		const u32	k
	)		[inline, static]

Initialize centroids.

• Input Arguments:
  • samples_in : Samples' matrix (i.e., each row is a sample).
  • k : k denotes the number of desired clusters.

• Output Arguments:
  • centroids_out : Centroids' matrix (i.e., each row is a centroid).

template<typename eT >

void op_kmeans::init_centroids	(	Mat< eT > &	centroids_out,
		const Mat< eT > &	samples_in,
		const u32	k
	)		[inline, static]

Initialize centroids.

• Input Arguments:
  • samples_in : Samples' matrix (i.e., each row is a sample).
  • k : k denotes the number of desired clusters.

• Output Arguments:
  • centroids_out : Centroids' matrix (i.e., each row is a centroid).