1.

What is Neuron Models and Discuss (Biological Neuron, Artificial Neuron, Mathematical
Model).
2.Explain about Multi-Layer Perceptron (MLP) Networks and the Error-Backpropagation
Algorithm.
3.what is Classification Decision Tree.? Explain Example of a Classification Decision Tree.
4. Explain Fisher’s Linear Discriminant and Thresholding for Classification.
5.Discuss about Strengths and Weaknesses of Decision-Tree Approach
6.Assume data set ID3 Decision Trees and calculate (example).
7.Define Perceptron? Explain the Perceptron Algorithm
8.Explain the Linear Maximal Margin Classifier for Linearly Separable Data.
9.Explain about Linear Discriminant Functions for Binary Classification.
10.Discuss about Kernel-Induced Feature Spaces.
11.Explain the Regression by Support Vector Machines.
12.what is Pruning the Tree? And Explain.

1.What is the primary goal of a Support Vector Machine (SVM) in classification? a.

Maximize margin between classes b. Minimize margin between classes c. Maximize error
rate d. Minimize training set accuracy
Answer: a.
2.What are support vectors in SVM? a. Data points that maximize margin b. Data points that
lie on the decision boundary c. Data points that minimize margin d. All data points in the
dataset
Answer: a.
3.Which of the following is a valid kernel function in SVM? a. Logistic b. Linear c.
Exponential d. Absolute value
Answer: b.
4.SVM is suitable for: a. Small datasets b. Large datasets c. Both small and large datasets d.
None of the above
Answer: c.
5.In SVM, the hyperplane with the maximum margin is chosen based on: a. Data points
closest to the hyperplane b. Data points in the middle of the margin c. All data points in the
dataset d. Randomly selected data points
Answer: a.
Neural Networks (NNs):
6.What is the basic unit of a neural network? a. Node b. Feature c. Hyperplane d. Decision
tree
Answer: a.
7.Which function introduces non-linearity in a neural network? a. Identity b. Linear c.
Sigmoid d. Exponential
Answer: c.
8.What is the purpose of the backpropagation algorithm in neural networks? a. To decrease
the number of hidden layers b. To optimize the hyperparameters c. To update weights and
biases d. To increase model complexity
Answer: c.
9.What does the term "deep" refer to in deep neural networks? a. Number of nodes b.
Number of layers c. Number of features d. Number of iterations
Answer: b.
10.Which type of neural network layer is responsible for reducing the dimensionality of the
input? a. Input layer b. Hidden layer c. Output layer d. Pooling layer
Answer: d.
11. ______ is a data compression technique in machine learning and search algorithms that
reduces the size of decision trees by removing sections of the tree that are non-critical and
redundant to classify instances.
Answer: c. a) pre pruning d)
post pruning c) pruning d) none of these
12. What are the measures of impurity for evaluating splits in decision trees.
a) gain ratio b) entropy reduction c) Gini index d) all of the above
Answer: d.
13. _____is a non-parametric supervised learning algorithm, which is utilized for both
classification and regression tasks. It has a hierarchical, tree structure, which consists of a
root node, branches, internal nodes and leaf nodes.
a) decision tree b) svm c) machine learning d) neural networks
Answer: a.
14. In Machine Learning, _______measures the level of disorder or uncertainty in a given
dataset or system. It is a metric that quantifies the amount of information in a dataset, and it is
commonly used to evaluate the quality of a model and its ability to make accurate
predictions.
a) entropy b) Gini index c) information gain d) all of these
Answer: a.
15. ____is the reduction in entropy or surprise by transforming a dataset and is often used in
training decision trees.
a) Gini index b) information gain c) entropy d) none of these
Answer: b.
16. ________ evaluates the input signals, determining the strength of each one; calculates a
total for the combined input signals and compares that total to some threshold level; and
determines what the output should be.
a) biological neuron b) artificial neuron c) neuron d) all of these
Answer: b.
17. Artificial neurons use an activation function, often called a transfer function, to compute
their activation as a function of total input stimulus.
a) true b) false c) none of these
Answer: a.
18. The number of input and output nodes needed for a network will depend on the nature of
the data presented to the network, and the type of the output desired from it, respectively is
a) Feed neural networks. b) Multi-layer perceptions c) Feedforward networks d) None
Answer: c.
19. ________ can be applied to problems, such as learning to categorize medical patients
according to their ailments/conditions, equipment defects according to their faults, loan
applications according to their payment failures, and so on.
a) decision tree learning b) neural learning c) deep learning d) none of these
Answer: a.
20_______is a heuristic for selection of the splitting criterion that best separates a given
dataset D of class-labelled training tuples into individual classes
a) entropy Measure b) statical Measure c) mathematical measure d) impurity measure
Answer: d.
21. _________is linked closely to a category of brain-development theories published by
cognitive neuro scientists in the early 1990s
a) deep learning b) machine learning c) artificial learning d) cognitive learning
Answer: a.
22. What is the primary goal of a Support Vector Machine (SVM) in classification?
a) Maximize margin between classes b) Minimize margin between classes c)
Maximize error rate d) Minimize training set accuracy
Answer: a.
23. The Least Mean Square (LMS) algorithm, originally formulated by____________
stochastic gradient algorithm that iterates weight vector w in the regressor in the direction of
the
a) Hoff and jack b) widrow and Hoff c) mike and Hoff d) side and widrow
Answer: b.
24. _______ provides a specific choice of direction for projection of n-dimensional data
down to one dimension.
a) fisher’s linear discriminant b) threshold
c) minimum description length principal d) all of these
Answer: a.
25.What are support vectors in SVM?
a) Data points that maximize margin b) Data points that lie on the decision boundary c) Data
points that minimize margin d) All data points in the dataset
Answer: a.
26. Support vector regression is the natural extension of methods used for
a) classification b) regression c) Bayesian perspective d) all of these
Answer: a.
27.Who proposed the machine—the perceptron—whose architecture encodes the structure of
a linear discriminant function
a) jack and widrow b) widrow and Hoff c)Rosenblatt d)none of these
Answer: c.
28.The_________algorithm tests the decision function g(x) on each element in the training
set, and if the test fails, it adjusts the free parameters w and w0 incrementally.
a) incremental b) perceptron c) error-square d) all of these
Answer: b.
29. _______Builds the first a fixed nonlinear mapping transforms the data into a new feature
space, and then a linear machine is used to classify the data in the new feature space
Answer: b.

a) linear machines b) nonlinear machines c) support vector machines d) all of these

30. In _______ typically some measure for error of approximation is used instead of margin
between an optimal separating hyperplane and support vectors, which was used in the design
of SV classifiers.

a) non regression b) regression c) classification d) support vector machines

Answer: b or d.

General:
31. In machine learning, when might SVMs be preferred over neural networks? a. When
interpretability is crucial b. When working with images c. When the dataset is large d. Both a
and c
Answer: d.
32. What is a disadvantage of neural networks compared to SVMs? a. They are less prone
to overfitting b. They require more data c. They are simpler to interpret d. They are faster to
train
Answer: b
33. Which of the following is a common use case for SVMs? a. Image classification b.
Natural language processing c. Anomaly detection d. Speech recognition
Answer: c.
34. What is the purpose of cross-validation in machine learning? a. To train a model on
multiple datasets b. To optimize hyperparameters c. To validate model performance on a
single dataset d. To replace the test set
Answer: b
35. Which algorithm is sensitive to outliers in the training data? a. SVM b. Neural
network c. Both d. Neither
Answer: a.
36. What is the activation function commonly used in the output layer of a binary
classification neural network? a. ReLU b. Sigmoid c. Tanh d. Softmax
Answer: b.
37. What does the term "overfitting" mean in the context of machine learning? a. The
model is too simple and cannot capture patterns in the data. b. The model performs well on
the training set but poorly on new data. c. The model is too complex and fits the noise in the
training data. d. The model is unable to converge during training.
Answer: c.
38. What is the primary advantage of using a radial basis function (RBF) kernel in SVM?
a. It is computationally efficient. b. It allows the model to handle non-linear relationships. c.
It reduces the risk of overfitting. d. It simplifies the interpretability of the model.
Answer: b.
39. Which of the following statements is true about neural networks? a. They are always
interpretable. b. They require fewer computational resources compared to SVMs. c. They
automatically learn hierarchical representations from data. d. They are not suitable for tasks
with non-linear relationships.
Answer: c.
40. In neural networks, what is the purpose of the activation function? a. To introduce
non-linearity b. To reduce the number of nodes c. To control the learning rate d. To
determine the kernel type
Answer: a.
41. What is the primary disadvantage of using a linear kernel in SVM? a. It cannot handle
non-linear relationships. b. It is computationally expensive. c. It is prone to overfitting. d. It
requires more training data.
Answer: a.
42. Which hyperparameter in SVM determines the trade-off between achieving a smooth
decision boundary and classifying training points correctly? a. C b. Gamma c. Kernel d.
Margin
Answer: a. C
43. In neural networks, what is the purpose of the validation set during training? a. To
train the model b. To fine-tune hyperparameters c. To evaluate model performance on unseen
data d. To test the model's accuracy
Answer: b
44. What is the primary advantage of using a non-linear activation function in a neural
network? a. It simplifies the model architecture. b. It allows the network to learn complex
patterns. c. It reduces the risk of underfitting. d. It speeds up the training process.
Answer: b.
45. What is the purpose of regularization in machine learning models? a. To increase
model complexity b. To decrease the learning rate c. To penalize overly complex models d.
To reduce the number of iterations
Answer: c.
46. Which of the following is a characteristic of a well-regularized machine learning
model? a. High training accuracy, low test accuracy b. Low training accuracy, low test
accuracy c. High training accuracy, high test accuracy d. Low training accuracy, high test
accuracy
Answer: c. High training accuracy, high test accuracy
47. In a neural network, what is the purpose of the dropout layer? a. To randomly remove
nodes during training b. To increase the number of hidden layers c. To reduce the learning
rate d. To enforce weight constraints
Answer: a. To randomly remove nodes during training
48. Which of the following is a disadvantage of using a polynomial kernel in SVM? a. It
is computationally expensive. b. It cannot handle non-linear relationships. c. It is prone to
overfitting. d. It may lead to high-dimensional feature spaces.
Answer: d. It may lead to high-dimensional feature spaces.
49. What is the role of the bias term in a neural network? a. It controls the learning rate. b.
It shifts the decision boundary. c. It reduces the number of hidden layers. d. It increases
model complexity.
Answer: b. It shifts the decision boundary.
50. In SVM, what does the term "soft margin" refer to? a. A margin that is too narrow b.
A margin that allows for some misclassification c. A margin that is too wide d. A margin that
is fixed and cannot be adjusted
Answer: b. A margin that allows for some misclassification
51. What is the purpose of the rectified linear unit (ReLU) activation function in neural
networks? a. To introduce non-linearity b. To enforce weight constraints c. To control the
learning rate d. To reduce the risk of overfitting
Answer: a. To introduce non-linearity
52. Which of the following is a characteristic of underfitting in machine learning models?
a. High training accuracy, high test accuracy b. Low training accuracy, high test accuracy c.
High training accuracy, low test accuracy d. Low training accuracy, low test accuracy
Answer: b. Low training accuracy, high test accuracy
53. In SVM, what does the term "kernel trick" refer to? a. A technique to reduce the
dimensionality of the feature space b. A method to handle non-linear relationships by
implicitly mapping data to a higher-dimensional space c. A strategy to minimize the margin
between classes d. A way to speed up the training process
Answer: b. A method to handle non-linear relationships by implicitly mapping data to a
higher-dimensional space
54. Which of the following is true about the bias-variance tradeoff in machine learning? a.
High bias leads to overfitting. b. High variance leads to underfitting. c. Both high bias and
high variance are desirable. d. It is the trade-off between model complexity and
generalization.
Answer: d. It is the trade-off between model complexity and generalization.
55. What is the purpose of cross-entropy loss in neural networks? a. To minimize the
mean squared error b. To maximize the margin between classes c. To penalize the model for
incorrect predictions d. To speed up the training process
Answer: c. To penalize the model for incorrect predictions
56. Which of the following is a common approach to prevent overfitting in neural
networks? a. Increasing the learning rate b. Adding more hidden layers c. Adding dropout
layers d. Removing the activation function
Answer: c. Adding dropout layers
57. What is the purpose of the hinge loss function in SVM? a. To minimize classification
error b. To maximize margin between classes c. To reduce the learning rate d. To enforce
weight constraints
Answer: b. To maximize margin between classes
58. What is the main advantage of using a Gaussian radial basis function (RBF) kernel in
SVM? a. It reduces the risk of overfitting. b. It allows the model to handle non-linear
relationships. c. It simplifies the interpretability of the model. d. It speeds up the training
process.
Answer: b. It allows the model to handle non-linear relationships.
59. Which of the following is a common activation function in the hidden layers of a
neural network? a. Sigmoid b. ReLU c. Softmax d. Tanh
Answer: b. ReLU
60. In SVM, how does the regularization parameter C affect the decision boundary? a.
Higher C values lead to a smoother decision boundary. b. Higher C values lead to a more
complex decision boundary. c. Lower C values lead to a wider margin. d. Lower C values
lead to a narrower margin.
Answer: b. Higher C values lead to a more complex decision boundary.

61. What is machine learning? ( )

A. ML is subset of Artificial intelligence B. ML is subset of deep learning
C. ML is subset of supervised D. ML is subset of unsupervised

62.Application of Machine learning is __________. ( )

A. Email filtering B. Sentimental analysis C. Face recognition D. All of the above

63. you'll know what data is best to use to train and test your model with ( )
A. Domain knowledge B. declarative knowledge
C. procedural knowledge D. integrated knowledge
64.which model use in Computational Learning Theory. ( )
A. PAC model B. Linear model C. Logical model D. Dtree model

65.Data set divided into ( )

A. 1 type B. 2 types C.3 types D. 4 types

66.which algorithms using in regression. ( )

A. Linear B. Dtree C. Single linear D. All

67.Application of machine learning methods to large databases is called ( )

A. Data mining. B. Artificial intelligence C.big data computing D.internet of things
68.In what type of learning labelled training data is used ( )
A. Unsupervised learning B. Supervised learning
C. Reinforcement learning D. Active learning
69. Following are the types of supervised learning. ( )
A. Regression B. Classification C. Association rules D. A & B
70.Type of matrix decomposition model is ( )
A. Descriptive model B. Predictive model C. Logical model D. None of the above
71. Supervised learning and unsupervised clustering both require which is correct according
to the statement ( )
A. Output attribute B. Hidden attribute C. input attribute D. categorical attribute
72. summarizing and organizing your data ( )
A. Descriptive Statistics B. Descriptive Statistics Analysis
C. Inferential statistics D. Inferential statistics Analysis
73. produced by more complex mathematical calculations, and allow us to infer trends and
make assumptions and predictions about a population based on a study of a sample taken
from it. ( )
A. Descriptive Statistics B. Descriptive Statistics Analysis
C. Inferential statistics D. Inferential statistics Analysis
74. Our model will not be trained well with the training data. There will be high training error
when we train our model with the data. ( )
A. Bias B. Variance C. Bias error D. Variance error
75.If you train your data on training data and obtain a very low error, upon changing the data
and then training the same previous model you experience high error
A. Bias B. Variance C. Bias error D. Variance error
76. ( )

High Training error

Bias - High
High Test error

Variance – High

A. Over Fitting B. Generalized Model

C.Under Fitting D.All

77.
( )
degree of polynomial = 2
Low Training error
Bias – Low
Low Test error
Variance – Low
A. Over Fitting B. Generalized Model

C.Under Fitting D.All

78. ( )

degree of polynomial = 3

Low Training error

Bias - Low

High Test error

Variance – High
A. Over Fitting B. Generalized Model

C.Under Fitting D.All

79.The hypothesis space defines the set of all possible models that can
be learned by the algorithm ( )
A. Set B. Model C. Value D. Process

80.What is a kind of algorithms that is able to convert weak learners to strong learners?
 ( )
A. Bagging B. Ada boost C. Boosting D. Xg boost