Machine Learning | Regularization - Lasso, Ridge, and OLS Regression | L1, L2 Regularizations
Regularization in Linear Models as used in machine learning helps attain the best balance between overfitting vs. underfitting and model complexity.
To attain this balance an optimal bias-variance tradeoff (also known as bias-variance decomposition) is required.
This video starts with a brief on regularization as used in Machine Learning, especially in the context of Linear models for classification and regression, with or without the use of Basis functions.
It then gives a conceptual brief on different types of losses, norms, and regularization.
The video then talks in detail about OLS, LASSO, and RIDGE regressions, and L1 Regularization and L2 regularization, explaining the difference between each and a graphical understanding of how LASSO regression leads to the SPARSE FEATURE matrix.
Link to linked video:
ML | Bias Variance Trade-Off:
• Machine Learning | Bias Variance Trade-Off
ML | Decision Trees:
• Machine Learning | Decision Trees - ID3, C...