6
$\begingroup$

I am aware that an ensemble machine learning model is a stack of two or more machine learning models. Is there a word to refer to those individual models that go into the ensemble model? (i.e. a specific term/jargon?)

I wish to refer to them and am not aware of any specific terminology to describe them. I've considered:

  • "sub-models"
  • "component models"
  • even "pancake models" (since people generally understand what's being referring to when they hear 'one of the pancakes in the model')
Improve this question
$\endgroup$
5
  • $\begingroup$ Welcome to the site! I think it is good to call them as individual models, because even if we don't apply ensemble models it does give some output just to maximize the best out of them we ensemble but calling them sub or component is not correct I feel. $\endgroup$ – Toros91 Apr 19 '18 at 3:32
  • $\begingroup$ in my opinion, individual models in the model esemble, is a very weak component, and that is a small part in esemble mode. but in solving algorithms used such as bosting, bagging, stacking, most of the bayes algorithm is a weak individual classification of the model $\endgroup$ – Anggi Yuniar Putri May 11 '18 at 12:09
  • $\begingroup$ in my opinion, the combination model, where the ensemble method performs the processing of weak data to be combined into a single value. example bagging, random forest, boosting. $\endgroup$ – Alvurqoni Ariven Yutzky May 11 '18 at 12:58
  • $\begingroup$ the Ensemble method is one of the Machine Learning techniques that combines several basic models to produce an optimal model. Because of this the Ensemble method is most popular among the world-renowned competition related to Machine Learning such as Netflix and Kaggle. The Ensemble method is also a form of incorporation of machine learning algorithms into a predictive model for reducing Bagging, Boosting and Stacking $\endgroup$ – Rizka Rofiq May 11 '18 at 13:30
  • $\begingroup$ as i know, there is no word to refer to those individual models that go into the ensemble model, but there is some techniques of ensemble methods that are used for Bagging, Boosting and Stacking. Which is have adventages one of them is to capture linear and simple as well non-linear complex relationships in the data. This can be done by using two different models and forming an ensemble of two. $\endgroup$ – Ari fahrezi May 11 '18 at 14:20
5
$\begingroup$

I am not aware of a specific definition. Wikipedia does not mention such a term either. I would prefer "components", "individual/constituent models", or something like that.

If you definitely want to find a "correct" term, one way to discover it (if it exists) is to look into an ensemble learning early paper. To arrive at one, a good way is to search for a survey paper (e.g. here) and go on a journey to the past to find that original study.

From a quick skim of the above, you might want to look at Exploratory Data Analysis (Tukey 77), if you manage to find the pdf.

Improve this answer
$\endgroup$
3
$\begingroup$

I have heard people calling them "weak learners" many times, but this is only when they are not very acurate themselves.

Improve this answer
$\endgroup$
1
  • $\begingroup$ The term 'weak learner' traditionally has a stricter sense. See this post $\endgroup$ – npit Apr 20 '18 at 0:12
3
$\begingroup$

It depends on the ensemble model technique. If you are going to use a bagging approach then the term individual models or alternative models is the appropriate. But in the case of boosting methods the appropriate term is weak learner (classifier). Boosting algorithms is a family of machine learning algorithms that convert weak learners to strong ones. A weak learner is defined to be a classifier that is only slightly correlated with the true classification (it can label examples better than random guessing).

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.