Python API

class exemcl.ExemplarClustering
Members

models

__init__(self: exemcl.ExemplarClustering, ground_set: numpy.ndarray[numpy.float64[m, n]], precision: str = 'fp32', device: str = 'gpu', worker_count: int = - 1) → None

Initializes the submodular function of Exemplar-based clustering with the following parameters:

Parameters

  • ground_set (int) – The ground set for the function (usually denoted as \(V\)).

  • precision (int) – Required floating point precision (possible values: fp16, fp32 or fp64).

  • device (int) – Computing device to use for function evaluation (possible values: gpu or cpu). Please keep in mind, that FP16 precision is not available with CPUs.

  • worker_count (int) – Number of parallel workers to consider (-1 defaults to all available cores).

__call__(S)

Evaluates the function value for a single set \(S\).

Parameters

S (ndarray) – Input data set \(S\) represented as data matrix with shape [n, d].

Returns

Function value \(f(S)\).

__call__(S_multi)

Evaluates the marginal function values for a set of sets \(\left\lbrace S_1, \dots, S_n \right\rbrace\).

Parameters

S_multi (List[ndarray]) – Input data sets \(\left\lbrace S_1, \dots, S_n \right\rbrace\) represented as data matrices with shape [n_i, d] for each \(S_i\).

Returns

Function values \(\left\lbrace f(S_1), \dots, f(S_n) \right\rbrace\).

__call__(S, e)

Evaluates the marginal function value for a single set \(S\) and a marginal element \(e\).

Parameters

  • S (ndarray) – Input data set \(S\) represented as data matrix with shape [n, d].

  • e (ndarray) – Input data vector \(e\) with shape [d, 1].

Returns

Marginal function value \(f(S \mid e)\).

__call__(S, e_multi)

Evaluates the marginal function value for a single set \(S\) and a set of marginal elements \(\left\lbrace e_1, \dots, e_n \right\rbrace\).

Parameters

  • S (ndarray) – Input data set \(S\) represented as data matrix with shape [n, d].

  • e_multi (List[ndarray]) – Input data vectors \(\left\lbrace e_1, \dots, e_n \right\rbrace\) with shape [d, 1] each.

Returns

Marginal function values \(\left\lbrace f(S \mid e_1), \dots, f(S \mid e_n) \right\rbrace\).

__call__(S_multi, e)

Evaluates the marginal function values for a set of sets \(\left\lbrace S_1, \dots, S_n \right\rbrace\) and a marginal element \(e\).

Parameters

  • S_multi (List[ndarray]) – Input data sets \(\left\lbrace S_1, \dots, S_n \right\rbrace\) represented as data matrices with shape [n_i, d] for each \(S_i\).

  • e (ndarray) – Input data vector \(e\) with shape [d, 1].

Returns

Marginal function values \(\left\lbrace f(S_1 \mid e), \dots, f(S_n \mid e) \right\rbrace\).