Implementation:Scikit learn Scikit learn BernoulliRBM

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

Overview

Concrete tool for Bernoulli Restricted Boltzmann Machine (RBM) feature extraction provided by scikit-learn.

Description

BernoulliRBM is a Restricted Boltzmann Machine with binary visible units and binary hidden units. Parameters are estimated using Stochastic Maximum Likelihood (SML), also known as Persistent Contrastive Divergence (PCD). It functions as an unsupervised feature extractor (transformer) that learns a nonlinear representation of the input data. The time complexity is O(d^2) assuming d is approximately n_features or n_components.

Usage

Use BernoulliRBM as a feature extraction step in a pipeline, particularly for binary or binarized data. It is commonly combined with a classifier in a pipeline for tasks like digit recognition. The learned hidden representations can capture complex patterns in the input data.

Code Reference

Source Location

• Repository: scikit-learn
• File: sklearn/neural_network/_rbm.py

Signature

class BernoulliRBM(ClassNamePrefixFeaturesOutMixin, TransformerMixin, BaseEstimator):
    def __init__(
        self,
        n_components=256,
        *,
        learning_rate=0.1,
        batch_size=10,
        n_iter=10,
        verbose=0,
        random_state=None,
    ):

Import

from sklearn.neural_network import BernoulliRBM

I/O Contract

Inputs

Name	Type	Required	Description
n_components	int	No	Number of binary hidden units (default=256)
learning_rate	float	No	Learning rate for weight updates (default=0.1)
batch_size	int	No	Number of examples per minibatch (default=10)
n_iter	int	No	Number of iterations/sweeps over the training dataset (default=10)
verbose	int	No	Verbosity level (default=0)
random_state	int, RandomState, or None	No	Random state for reproducibility

Outputs

Name	Type	Description
intercept_hidden_	ndarray of shape (n_components,)	Biases of the hidden units
intercept_visible_	ndarray of shape (n_features,)	Biases of the visible units
components_	ndarray of shape (n_components, n_features)	Weight matrix connecting visible and hidden units
h_samples_	ndarray of shape (batch_size, n_components)	Hidden activation sampled from the model distribution
n_features_in_	int	Number of features seen during fit

Usage Examples

Basic Usage

from sklearn.neural_network import BernoulliRBM
from sklearn.pipeline import Pipeline
from sklearn.linear_model import LogisticRegression
from sklearn.datasets import load_digits
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import MinMaxScaler

X, y = load_digits(return_X_y=True)
X = MinMaxScaler().fit_transform(X)
X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=42)

rbm = BernoulliRBM(n_components=100, learning_rate=0.06, n_iter=20, random_state=42)
logistic = LogisticRegression(max_iter=1000)

pipe = Pipeline([("rbm", rbm), ("logistic", logistic)])
pipe.fit(X_train, y_train)
print(f"Accuracy: {pipe.score(X_test, y_test):.3f}")

Related Pages

• Principle:Scikit_learn_Scikit_learn_Neural_Networks