Precision Recall

Module: benchmarks.metrics.precision_recall

Source: benchmarks/metrics/precision_recall.py

Overview

Precision-recall metrics for evaluating generative models.

This module implements precision and recall metrics for generative models as described in "Improved Precision and Recall Metric for Assessing Generative Models" (Kynkäänniemi et al., 2019).

The implementation uses clustering to identify modes in the data distribution and computes precision and recall based on cluster coverage.

Classes

KMeansModule

class KMeansModule

PrecisionRecallBenchmark

class PrecisionRecallBenchmark

Functions

init

def __init__()

init

def __init__()

compute_cluster_based_metrics

def compute_cluster_based_metrics()

compute_distance_based_metrics

def compute_distance_based_metrics()

compute_precision_recall

def compute_precision_recall()

fit

def fit()

is_well_separated_clusters

def is_well_separated_clusters()

run

def run()

Module Statistics

• Classes: 2
• Functions: 8
• Imports: 5