Which model type directly maps input features to class labels in machine learning?

The discriminative model is designed to directly map input features to class labels in machine learning tasks. By focusing on modeling the decision boundary between different classes, these models learn how to distinguish between the various classes given the input data. The primary objective is to estimate the conditional probability of the class labels given the input features, allowing for effective classification.

This approach is particularly powerful for supervised learning tasks where labeled data is available, enabling the model to directly learn from the relationships between input data and the corresponding outputs. Examples of discriminative models include logistic regression, support vector machines, and neural networks, which all aim to find patterns within the data that correlate with specific class labels.

In contrast, generative models focus on modeling the distribution of individual classes and can generate new samples from the learned distribution. Predictive models, while closely related to the concept of mapping inputs to outputs, do not necessarily encapsulate the distinction in how the decision boundaries are learned. Hybrid models combine characteristics from both generative and discriminative models, which may not directly reflect the specific focus of mapping input features to class labels as clearly as discriminative models do.