Zettelkasten/Terminology Information
Ensemble learning
Computer-Nerd
2023. 3. 15. 12:29
Information
- Ensemble learning combines the predictions of multiple models to achieve better accuracy than a single model.
- There are three types of ensemble learning: bagging, boosting, and stacking.
- Bagging (bootstrap aggregating) involves training multiple models on different subsets of the training data and then averaging their predictions.
- Boosting involves training multiple models in sequence, with each model focusing on the examples that the previous model misclassified.
- Stacking is to use several different models to make predictions on a set of data, and then use the outputs of these models as input features to a meta-model, which makes the final prediction.
- Ensemble learning can help reduce overfitting, improve generalization, and increase model robustness.
- However, it can also increase model complexity, training time, and computational resources required.
- Ensemble learning is widely used in a variety of machine learning applications, including image and speech recognition, natural language processing, and recommender systems.