Random patches

As explained in Ensemble Methods for Machine Learning

Random patches is an ensemble technique that trains models using independently sampled subsets of both training examples and input features.

Imagine building many image classifiers, each seeing a different set of images and pixels. Random patches creates this diversity by independently sampling both training examples and input features for every base estimator. section 2.4.2 section 2.0

Build a random-patches ensemble procedurally:

• Choose a base learner, such as an SVC; the method is not restricted to decision trees. section 2.4.2
• For each estimator, sample a subset of training examples and a subset of features, with or without replacement. section 2.4.2
• Train the learner on its sampled rows and columns, then aggregate predictions across the independently trained estimators. section 2.0 section 2.6

A scikit-learn configuration

Use BaggingClassifier with max_samples=0.75 and bootstrap=True for example sampling, plus max_features=0.5 and bootstrap_features=True for feature sampling. Sampling features can make learners more efficient on image or text data with many dimensions. section 2.4.2

Figure 2.9 Bagging compared to random subspaces and random patches. The unshaded rows and columns represent training examples and features, respectively, that have been left out.

Compare which rows and columns remain available to bagging, random subspaces, and random patches. section 2.4.2

Unlike random forests, random patches samples features once before each base learner runs; random forests sample features inside randomized decision-tree growth. section 2.4.2

How Ensemble Methods for Machine Learning covers it

How deep it goes

Explains row-and-column sampling, estimator choice, implementation parameters, and contrast with random forests.

Examples

Uses image and text databases, SVC learners, and BaggingClassifier configurations with row and feature sampling.

What it assumes

Assumes ensemble learning, base estimators, bagging, and model aggregation while reinforcing sampling-based diversity.

What you can do after

Apply random patches through BaggingClassifier or another base learner to increase diversity and smooth predictions.

How it connects

Connects bagging, random subspaces, random forests, ensemble diversity, model aggregation, and support vector machines.

liveBookmanning.com

Read Ensemble Methods for Machine Learning on liveBook Get the book at manning.com

Also explained in

• 

  liveBookmanning.com
  Machine Learning for Tabular Data: XGBoost, Deep Learning, and AIRandom patches is an ensemble sampling technique that trains models on randomly selected subsets of both observations and features.

Compare how 2 books explain random patches →

Related concepts

• baggingkind of · Bagging, or bootstrap aggregation, is an ensemble technique that trains multiple models on bootstrap samples of data and combines their predictions.
• random subspacesuses · Random subspaces is an ensemble technique that trains each model on a randomly selected subset of input features.

This explainer was generated by Manning's knowledge graph from the text of Ensemble Methods for Machine Learning. The section links point to the parts of the book it draws on.