Cognitive Scaffold

Preparing your thinking workspace

arrow_back_ios_new
MENTAL MODEL · M6030

Regularization

Regularization
Learn & MetacognitionHigh supportLearning Science
Included
account_tree

Version 1.0.0 · Updated 2026-07-30

CORE DEFINITION

Add a penalty term (such as the absolute value of parameters) to the loss function to limit model complexity and prevent overfitting. It is better to sacrifice a little training accuracy to keep the model simple. Scaffolding role: artificially introduced constraints. When interpreting the world or making plans, if you find the explanation too complex (overfitting), force yourself to eliminate secondary variables by introducing a "regularization term" (Occam's razor). Simple models often generalize better.

SCAFFOLDING EFFECT

psychology

Reduce cognitive load

Artificially introduced constraints. When interpreting the world or making plans, if you find the explanation too complex (overfitting), force yourself to eliminate secondary variables by introducing a "regularization term" (Occam's razor). Simple models often generalize better.

anchor

Anchor fast decisions

Add a penalty term (L1/L2) to the loss to constrain parameter scale, reduce model complexity, improve generalization, and suppress overfitting.

MINIMUM ACTION

In progress 0/4

Practice this model in one real situation:

Check to track your progress (stored locally)
Learning progress0%
account_treeGenealogyexpand_more
menu_bookReferencesexpand_more

Source support: Explicit

  • link
    en.wikipedia.orghttps://en.wikipedia.org/wiki/RegularizationZH · Explicit
    verified

RELATED MODELS