Artificial Intelligence Flashcards
In Bayesian parameter estimation, the ________ distribution is often used as a prior or a real-valued quantity
Gaussian
In a Bayes classifier with Gaussian class-conditional densities, the parameters that need to be estimated for each class are the _____ and ________ _______ of the Gaussian distribution
mean, covariance matrix
The core principle of _________ statistics is using prior knowledge and observed data to update beliefs.
Bayesian
MAP stands for ________ ___ _______ in the context of Bayesian estimation.
Maximum A Posteriori
The property of Gaussian distribution that is NOT directly related to its use in the central limit theorem is the highest entropy among distributions with a given mean and variance. (True / False)
True
The ________ ______ theorem states that sums of independent random variables have a Gaussian distribution.
central limit
Kschischang’s Algorithm is proposed for message passing on a factor graph in _________ ________ __________.
Loopy Belief Propagation
The role of a regularization term in a machine learning model is to reduce overfitting by penalizing large coefficients. (True / False)
True
Training data is NOT typically a component of a Bayesian network. (True / False)
True
In _______ learning, the goal is to learn a mapping from input ( X ) to output ( Y ) based on a training set of input-output pairs. This process is primarily focused on prediction.
supervised
The Distribution rule is a basic rule of probability. (True / False)
False
The Distribution rule is NOT a basic rule of probability.
Bayes’ theorem is a crucial concept in probabilistic machine learning. It relates the conditional and marginal probabilities of random variables.
What is the formula for Bayes’ theorem?
(P|A)= [P(B|A) P(A)]/P(B)
A Gaussian distribution is also known as ______ ________.
Normal distribution
The parameters that fully define a univariate Gaussian (normal) distribution are the ________ and the __________.
mean (μ), variance (σ²)
The _______ distribution represents the updated belief about a parameter after observing data in Bayesian inference.
posterior
Gradient Descent is a common method for approximating posterior distributions in Bayesian inference. (True / False)
False
Gradient Descent is NOT a common method for approximating posterior distributions in Bayesian inference.
The conjugate prior for the Gaussian likelihood with known variance is the _________ ____________.
Normal distribution
In the context of conjugate priors, a prior is termed “________” if the posterior distribution belongs to the same family of distributions as the prior distribution.
conjugate
___________ __________ __________ is commonly used to estimate the parameters in linear regression.
Maximum Likelihood Estimation (MLE)
In linear regression, the assumption NOT typically made about the errors (residuals) is that they are equal to zero. (True / False)
True
K-means is typically used for classification tasks. (True / False)
False
K-means is NOT typically used for classification tasks.
In the context of logistic regression, the ________ function maps predicted values to probabilities.
sigmoid
In Hidden Markov Models (HMMs), the ‘_____’ states represent the underlying system states that are not directly observed.
hidden
