Font size
WorksheetsPage 1
Total questions: 150
Worksheet time: 1hrs 15mins
Machine Learning is a field that enables computers to learn without being
Programmed
Explicitly programmed
Supervised
Controlled
Learning from labeled data is called
Unsupervised learning
Supervised learning
Reinforcement learning
Deep learning
Candidate Elimination algorithm is used in
Clustering
Concept learning
Regression
Reinforcement
Perceptron is a
Linear classifier
Clustering algorithm
Regression model
Optimization method
Linear regression is used for
Classification
Clustering
Prediction of continuous values
Feature selection
Version space represents
All possible hypotheses
Training data
Testing data
Output values
Perceptron works only when data is
Non-linear
Linearly separable
Random
Unlabeled
Gradient descent is used to minimize
Accuracy
Error
Clusters
Data size
Concept learning is treated as a
Search problem
Clustering problem
Regression problem
Optimization problem
Training data contains
Only inputs
Only outputs
Input-output pairs
Random values
Linear discriminants are used for
Clustering
Classification
Regression
Optimization
Which is supervised learning?
K-means
PCA
Decision tree
Apriori
Perceptron learning updates
Weights
Inputs
Outputs
Labels
Hypothesis space contains
All possible models
Training samples
Errors
Outputs
Overfitting occurs when model performs well on
Training data only
Test data only
New data
All data
Learning from labeled data is called (a) learning.
Perceptron is a (a) classifier.
Candidate Elimination algorithm maintains (a) sets.
Linear regression predicts (a) values.
Concept learning is viewed as a (a) problem.
Version space lies between (a) and specific hypotheses.
Perceptron works on (a) separable data.
Training data contains input and (a) .
Gradient descent minimizes (a) .
Supervised learning requires (a) data.
Hypothesis space is a set of (a) .
Classification predicts (a) labels.
Regression predicts (a) outputs.
Linear discriminants are used for (a) .
Perceptron is a single layer (a) network.
MLP stands for
Multi Layer Perceptron
Maximum Likelihood Predictor
Multi Level Processor
Modular Learning Protocol
Backpropagation is used to
initialize random weights
update weights
compute input features
normalize outputs
RBF stands for
Rectified Bilinear Function
Random Basis Filter
Radial Basis Function
Residual Backward Flow
SVM is mainly used for
clustering
regression only
classification
dimensionality reduction
Error in neural networks is propagated
forward
laterally
backward
randomly
Kernel trick is used in
PCA
K-means
SVM
Naive Bayes
Hidden layers are present in
single-layer perceptron
logistic regression
MLP
k-NN
Curse of dimensionality mainly affects (a) .
Activation functions are used in (a) .
SVM maximizes (a) .
Sigmoid is an (a) .
Training in MLP is (a) .
RBF networks use (a) .
Hyperplane is a concept in (a) .
Backpropagation minimizes (a) .
Backpropagation updates (a) .
MLP contains input, hidden and (a) layers.
SVM finds optimal (a) .
RBF networks use (a) functions.
Kernel functions map data to (a) dimensions.
Learning in neural networks is (a) .
Error is propagated (a) .
SVM is a (a) classifier.
Hidden neurons use (a) functions.
Training data is required for (a) learning.
Overfitting can occur in (a) networks.
RBF network is a type of (a) network.
SVM can handle (a) separable data.
Backpropagation uses (a) descent.
Output of neuron depends on (a) .
Decision trees are used for
clustering only
classification only
regression only
classification and regression
Entropy measures
purity
impurity
margin
variance
Information gain is based on
entropy
Gini index
variance reduction
margin maximization
CART stands for
Classification and Regression Trees
Cluster Analysis and Rule Trees
Conditional Attribute Ranking Trees
Combined and Averaged Regression Techniques
Bagging is an
optimization algorithm
ensemble method
feature selection technique
distance metric
Boosting combines
weak learners
strong learners
random features
bootstrap samples
K-means is
supervised learning
unsupervised learning
reinforcement learning
semi-supervised learning
Random Forest uses
boosting
stacking
bagging
pruning
Gini index measures
variance
purity
impurity
entropy
Pruning is used to reduce
training time
underfitting
overfitting
feature dimensionality
KNN is based on
probability
information gain
distance
gradient
Clustering uses
labeled data
unlabeled data
balanced data
oversampled data
Leaf node represents
root
split attribute
class label
entropy
Ensemble methods improve
training data size
accuracy
feature count
label count
GMM is used for
regression
classification only
clustering
feature scaling
Decision trees are used for (a) .
Entropy measures (a) .
Information gain selects the best (a) .
CART stands for (a) .
Combining models is called (a) learning.
Bagging reduces (a) .
Boosting reduces (a) .
K-means is an (a) algorithm.
Random Forest is based on (a) .
Pruning reduces (a) .
KNN uses (a) metric.
Leaf nodes give (a) .
Gini index measures (a) .
Clustering groups (a) data.
Gaussian Mixture Models are used for (a) .
PCA stands for
Principal Component Analysis
Partial Correlation Analysis
Probabilistic Component Aggregation
Principal Cluster Analysis
PCA is used for
hyperparameter tuning
dimensionality reduction
feature scaling
model selection
LDA stands for
Latent Dirichlet Allocation
Linear Discriminant Analysis
Local Density Approximation
Logistic Discriminant Analysis
ICA stands for
Independent Component Analysis
Integrated Cluster Analysis
Incremental Component Averaging
Independent Correlation Assessment
Isomap is a
linear classifier
non-linear reduction technique
clustering algorithm
feature selection method
LLE stands for
Locally Linear Embedding
Least Likelihood Estimation
Linear Laplacian Encoding
Localized Logistic Embedding
Genetic Algorithms belong to
instance-based learning
evolutionary learning
reinforcement learning
supervised learning
Crossover is a
gradient estimator
regularization trick
genetic operator
clustering heuristic
Complete the statement: Mutation introduces (a) .
Complete the statement: PCA works on (a) .
Complete the statement: LDA maximizes (a) .
Complete the statement: Dimensionality reduction removes (a) .
Complete the statement: Genetic algorithms are inspired by (a) .
Complete the statement: Optimization in ML uses (a) .
Complete the statement: Feature extraction is done by (a) .
Fill in the blank: PCA reduces (a) .
Fill in the blank: LDA is a (a) method.
Fill in the blank: ICA separates (a) signals.
Fill in the blank: LLE is a (a) technique.
Fill in the blank: Genetic algorithms use (a) operators.
Fill in the blank: Crossover and mutation are (a) operations.
Fill in the blank: PCA uses (a) of data.
Fill in the blank: Isomap preserves (a) distance.
Fill in the blank: Dimensionality reduction improves (a) .
Fill in the blank: Factor analysis is used for (a) .
Fill in the blank: Mutation introduces (a) .
Fill in the blank: Selection chooses (a) individuals.
Fill in the blank: Genetic algorithms are (a) search techniques.
Fill in the blank: PCA produces (a) components.
Fill in the blank: LDA maximizes (a) separation.
Choose the correct option: RL stands for ________.
Reinforcement Learning
Representation Learning
Relational Logic
Regression Learning
Choose the correct option: RL learns by ________.
trial and error
memorization of labels
deterministic planning
unsupervised clustering
Choose the correct option: MCMC stands for ________.
Markov Chain Monte Carlo
Multiple Constraint Monte Carlo
Markov Controlled Model Class
Matrix Chain Monte Computation
Choose the correct option: HMM stands for ________.
Hidden Markov Model
Hierarchical Mixture Model
Hybrid Markov Mechanism
Hidden Memory Machine
Choose the correct option: Bayesian networks are ________.
probabilistic models
deterministic rule sets
distance metrics
optimization solvers
Choose the correct option: RL uses ________.
reward signals
class labels
genetic crossover
kernel tricks
Choose the correct option: MCMC is used for ________.
sampling
feature scaling
graph layout
error correction
Choose the correct option: Graphical models represent ________.
probabilistic relations
geometric transforms
deterministic pipelines
sorting procedures
Choose the correct option: HMM is used for ________.
sequence modeling
image denoising
nonlinear control
spatial indexing
Choose the correct option: In reinforcement learning, a policy defines ________.
actions
transitions between labels
distance metrics
network topology
Choose the correct option: A Markov process depends on ________.
current state
entire history
next reward only
random seed
Choose the correct option: Tracking methods commonly use ________.
HMM
k-d trees
support vector regression
Fourier descriptors
Choose the correct option: Bayesian networks are ________.
directed graphs
undirected trees
fully connected meshes
acyclic lattices
Choose the NB: correct option: An RL agent interacts with ________.
environment
labeling function
static dataset only
kernel
Choose the correct option: Monte Carlo methods use ________.
random sampling
deterministic integration
grid search only
symbolic algebra
RL stands for (a) .
RL learns through (a) and error.
MCMC is used for (a) .
HMM stands for (a) .
Bayesian networks are (a) models.
Reward guides (a) learning.
Graphical models represent (a) .
Markov process depends on (a) state.
Policy maps states to (a) .
HMM is used for (a) data.
Tracking methods use (a) .
Monte Carlo uses (a) sampling.
RL agent interacts with (a) .
Proposal distribution is used in (a) .
Markov Random Fields are (a) models.
