SYSTEMA CONSTRUCTUM

Accepted ontology entry

gradient boosting

Gradient boosting is a supervised learning algorithm that constructs an ensemble of weak predictive models (typically decision trees) in a stage-wise additive fashion. Each iteration fits a new model to the residual errors (negative gradie…

ACCEPTED THINGcmsdquogo046i3vv3b9uozns3

Definition

Gradient boosting is a supervised learning algorithm that constructs an ensemble of weak predictive models (typically decision trees) in a stage-wise additive fashion. Each iteration fits a new model to the residual errors (negative gradients) of the current ensemble, using a differentiable loss function to guide the gradient descent in function space. The final prediction is the weighted sum of all base model outputs. Parameters include learning rate (step size), number of estimators, tree depth, subsampling ratio, and regularization terms (L1/L2 on leaf weights). Persistence mechanism: mathematical specification formalized by Friedman (2001), implemented in open-source libraries (XGBoost, LightGBM, scikit-learn), and embedded in data science pedagogy. [formal: gradus_boostare | substrate: behavior | horizon: hours | explicit: yes | epoch: 0.91]

Why it is in scope

A human-made ensemble machine learning method that builds models sequentially, where each new model corrects the residual errors of the combined previous models by fitting to the negative gradient of the loss function. Persisted through software implementations (XGBoost, LightGBM), mathematical formalism, and pedagogical use in predictive modeling.

Names and aliases

Relations from this entry

  • cmsddehj003sp3vv3r6h06pb8INSTANCE_OF →

    Gradient boosting IS a specific kind of statistical method: an ensemble technique that builds models sequentially, each correcting the residual errors of its predecessor. A competent speaker calls it a statistical method.

  • cmrg0scos00ef2a1nklfvbk7xINSTANCE_OF →

    Gradient boosting IS a specific kind of machine learning: it is an ensemble algorithm that builds models sequentially, with each new model correcting the residual errors of the previous ones. A competent speaker calls gradient boosting 'a machine learning algorithm.' Machine learning is the nearest kind.

Relations to this entry

No accepted relations in this direction.

Record identity

Created
Aug 3, 2026, 9:31 PM UTC
Content hash
3a89ed851c8cdb0ac0a0eca788611e9a9820bb790aa1603441fb948cdb978000

Open a related act record