frsutils.core.similarities Namespace Reference

Data Structures
class	GaussianSimilarity
class	LinearSimilarity
class	SimilarityFunction

Functions
np.ndarray	calculate_similarity_matrix (np.ndarray X, SimilarityFunction similarity_func, Callable[[np.ndarray, np.ndarray], np.ndarray] tnorm)

Detailed Description

Similarities

Provides an extensible and optimized framework to compute similarity matrices
with pluggable similarity functions (using inheritance)

@author: Mehran Amiri

Function Documentation

calculate_similarity_matrix()

np.ndarray frsutils.core.similarities.calculate_similarity_matrix	(	np.ndarray	X,
		SimilarityFunction	similarity_func,
		Callable[[np.ndarray, np.ndarray], np.ndarray]	tnorm )

Compute the pairwise similarity matrix for samples using a vectorized
similarity function and T-norm operator.

@param X: Normalized input matrix of shape (n_samples, n_features)
@param similarity_func: Instance of SimilarityFunction subclass
@param tnorm: Callable T-norm function (e.g., min, product, Yager)
@return: Similarity matrix of shape (n_samples, n_samples)

Definition at line 58 of file similarities.py.

) -> np.ndarray:
    """
    Compute the pairwise similarity matrix for samples using a vectorized
    similarity function and T-norm operator.
 
    @param X: Normalized input matrix of shape (n_samples, n_features)
    @param similarity_func: Instance of SimilarityFunction subclass
    @param tnorm: Callable T-norm function (e.g., min, product, Yager)
    @return: Similarity matrix of shape (n_samples, n_samples)
    """
    n_samples, n_features = X.shape
    sim_matrix = np.ones((n_samples, n_samples), dtype=np.float64)
 
    for k in range(n_features):
        col = X[:, k].reshape(-1, 1)  # Shape (n, 1)
        diff = col - col.T            # Shape (n, n)
        sim_k = similarity_func(diff) # Compute similarity
        sim_matrix = tnorm(sim_matrix, sim_k)  # Apply T-norm
 
    np.fill_diagonal(sim_matrix, 1.0)
    return sim_matrix