Probabilistic Discriminative Models Quiz

What is a key characteristic of probabilistic discriminative models?

• They model the joint distribution $p(x, y)$

• They model the conditional distribution $p(y \mid x)$ directly

• They do not use probabilities

• They only work for unsupervised learning

Which of the following is an example of a probabilistic discriminative model?

• Naive Bayes classifier

• Gaussian Mixture Model

• Logistic Regression

• Hidden Markov Model

In probabilistic discriminative models, what does the term “discriminative” imply?

• The model discriminates between classes by learning the prior probabilities.

• The model learns to directly distinguish between classes by modeling $p(y \mid x)$.

• The model generates data samples for each class.

• The model does not involve any probability distributions.

Which loss function is typically used to train probabilistic discriminative models like logistic regression?

• Mean Squared Error

• Hinge Loss

• Cross-Entropy Loss

• Absolute Error

Which statement is true about the difference between generative and discriminative models?

• Generative models model $p(y \mid x)$, discriminative models model $p(x, y)$.

• Discriminative models can generate new data points, generative models cannot.

• Generative models model $p(x, y)$, discriminative models model $p(y \mid x)$.

• Both model $p(y \mid x)$ in the same way.

In logistic regression, what function is applied to the linear combination of features to get probabilities?

• Sigmoid function

• ReLU function

• Softmax function

• Tanh function

Which method is typically used for parameter estimation in probabilistic discriminative models?

• Maximum Likelihood Estimation (MLE)

• Expectation-Maximization (EM)

• Bayesian Inference

• Principal Component Analysis (PCA)