Implementation:Scikit learn Scikit learn LabelPropagation

Knowledge Sources	Scikit_learn Scikit-learn Docs
Domains	Machine Learning, Semi-Supervised Learning
Last Updated	2026-02-08 15:00 GMT

Overview

Concrete tool for semi-supervised classification using label propagation algorithms provided by scikit-learn.

Description

This module implements two semi-supervised classification algorithms based on label propagation on graphs: LabelPropagation and LabelSpreading. Both algorithms construct a fully-connected graph between all data points and propagate label information from labeled to unlabeled points. LabelPropagation uses hard clamping (labeled points never change), while LabelSpreading uses soft clamping (labels can change by a fraction alpha per iteration). Both support RBF and KNN kernels, with KNN producing sparse matrices for better scalability.

Usage

Use LabelPropagation or LabelSpreading when you have a small amount of labeled data and a large amount of unlabeled data. LabelSpreading is generally more robust to noise due to its soft clamping approach.

Code Reference

Source Location

• Repository: scikit-learn
• File: sklearn/semi_supervised/_label_propagation.py

Signature

class BaseLabelPropagation(ClassifierMixin, BaseEstimator, metaclass=ABCMeta):
    """Base class for label propagation module."""
    ...

class LabelPropagation(BaseLabelPropagation):
    """Label Propagation classifier."""

    def __init__(
        self,
        kernel="rbf",
        *,
        gamma=20,
        n_neighbors=7,
        max_iter=1000,
        tol=1e-3,
        n_jobs=None,
    ):
        ...

class LabelSpreading(BaseLabelPropagation):
    """LabelSpreading model for semi-supervised learning."""

    def __init__(
        self,
        kernel="rbf",
        *,
        gamma=20,
        n_neighbors=7,
        alpha=0.2,
        max_iter=30,
        tol=1e-3,
        n_jobs=None,
    ):
        ...

Import

from sklearn.semi_supervised import LabelPropagation, LabelSpreading

I/O Contract

Inputs

Name	Type	Required	Description
X	array-like of shape (n_samples, n_features)	Yes	Training data (labeled and unlabeled)
y	array-like of shape (n_samples,)	Yes	Target labels; -1 indicates unlabeled samples
kernel	str or callable	No	Kernel function: 'rbf' or 'knn' (default: 'rbf')
gamma	float	No	RBF kernel parameter (default: 20)
n_neighbors	int	No	Number of neighbors for KNN kernel (default: 7)
alpha	float	No	Clamping factor for LabelSpreading (default: 0.2)
max_iter	int	No	Maximum iterations (default: 1000 for LP, 30 for LS)
tol	float	No	Convergence tolerance (default: 1e-3)

Outputs

Name	Type	Description
transduction_	ndarray of shape (n_samples,)	Predicted labels for all samples including unlabeled
label_distributions_	ndarray of shape (n_samples, n_classes)	Label probability distribution for each sample
classes_	ndarray of shape (n_classes,)	Unique class labels
n_iter_	int	Number of iterations run

Usage Examples

Basic Usage

import numpy as np
from sklearn.semi_supervised import LabelPropagation
from sklearn.datasets import load_iris

iris = load_iris()
rng = np.random.RandomState(42)

# Mask some labels as unlabeled (-1)
labels = np.copy(iris.target)
random_unlabeled = rng.rand(len(labels)) < 0.3
labels[random_unlabeled] = -1

# Fit label propagation
lp = LabelPropagation(kernel='rbf', gamma=20)
lp.fit(iris.data, labels)
print("Accuracy on all:", lp.score(iris.data, iris.target))

Related Pages

• Principle:Scikit_learn_Scikit_learn_Semi_Supervised_Learning