This chapter is from the book
This chapter is from the book
This chapter is from the book
2.9 Summary
To summarize the main points discussed in this chapter:
Supervised learning is the task of inferring a function from labeled training data that can accurately predict the output label for new, unseen inputs.
The main two types of supervised learning problems are classification, where the goal is to predict to which class a sample belongs, and regression, where the goal is to predict a continuous value for a target variable.
The model’s hypothesis is a function learned from the training data that approximates the true mapping from features to labels.
A loss function measures the error of the model’s predictions on the training data and is used to guide the optimization process of the model.
Generative models learn the joint probability distribution of inputs and labels, while discriminative models focus on learning the decision boundary between classes.
Two common methods for estimating model parameters are maximum likelihood estimation (MLE) and Bayesian inference. MLE seeks to find the parameter values that maximize the likelihood of the observed data, whereas Bayesian inference combines beliefs with the observed data to produce a posterior distribution over the parameters.
The bias–variance tradeoff represents the tension between a model’s ability to perform well on the training set (bias) and its ability to generalize to unseen data (variance).
Regularization is a common technique to combat overfitting by penalizing complex models.
A typical machine learning pipeline includes data collection and preparation, model selection, training, hyperparameter tuning, performance evaluation, and deployment.
Supervised machine learning involves navigating a complex landscape of challenges, ranging from data quality and quantity to model complexity and interpretability.
