https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.

com/company/pianalytix/

Cracking the Code: 200+

Interview MCQs
(on Data Science, Machine Learning and Deep Learning)

1. In machine learning, what term is used to describe the problem of a model performing well on the training
data but poorly o n uns een data?

A) Stochastic Gradient Descent (SGD)

B) Adam Optimizer

C) Levenberg-Marquardt Algorithm

2. Which of the following algorithms is best suited for anomaly detection in unstructured data like images or
audio?

A) K-means clustering

B) Support Vector Machines (SVM)

C) Autoencoders

3. In a neural network, what is the purpose of the activation function?

A) Normalize the input data

B) Regularize the model to prevent overfitting

C) Introduce non-linearity to the model

4. W hich evaluation metric is co mmon ly used for imbalanced classification problems to assess model
p e r f o r ma nce?

A) Mean Squared Error (MSE)

B) Receiver Operating Characteristic (ROC) AUC

C) R-squared (R2) score

5. In the context of Natural Language Processing (NLP), what does TF-IDF stand for?

A) Term Frequency-Inverse Document Frequency

B) Text Frequency-Inverse Data Filter

C) Total Frequency-Inverse Document Filtering

6. W hich technique is used to reduce the dimensionality of data while preserving its variance in multivariate
analysis?

A) Principal Component Analysis (PCA)

B) Support Vector Machine (SVM)

C) Total Frequency-Inverse Document Filtering

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

7. In Reinforcement Learning, wh a t is the n a me of the technique used to estimate the expected long-term
reward from a given state?

A) Q-learning

B) Policy Gradient

C) Value Function

8. W hich method is used for handling missing data by creating multiple imputations based on the observed
d a t a?

A) K-nearest neighbors (KNN)

B) Mean substitution

C) Multiple Imputation by Chained Equations (MICE)

9. Which algorithm is a variant of the R a n d o m Forest algorithm and uses gradient boosting for improved
p e r f o r man ce?

A) XGBoost

B) Naive Bayes

C) K-means

10. In deep learning, which layer type is responsible for reducing the spatial dimensions of the input data?

A) Dense layer

B) Convolutional layer

C) Pooling layer

11. W h ich optimization algorithm updates the model parameters with a learning rate that adapts based on
the historical gradients?

A) Stochastic Gradient Descent (SGD)

B) Adam Optimizer

C) Levenberg-Marquardt Algorithm

12. In a time series forecasting problem, what is the term used to describe the pattern where the data
exhibits upward or downw ard trends over a long period?

A) Seasonality

B) Cyclical pattern

C) Trend

13. W hich statistical test is used to determine if there is a significant relationship between two categorical
variables in a dataset?

A) Pearson correlation coefficient

B) Chi-Square test

C ) t-test

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

14. W h e n building a recommendation system, which method uses collaborative filtering based on user-item
interac tions?

A) Content-based filtering

B) Singular Value Decomposition (SVD)

C) Apriori algorithm

15. Which algorithm is commonly used for semi-supervised learning tasks, where the training data has both
labeled an d unlabeled examples?

A) K-nearest neighbors (KNN)

B) Decision Tree

C) Expectation-Maximization (EM)

16. In time series analysis, what is the purpose of differencing in the Box-Jenkins method?

A) Remove seasonality

B) Make the series stationary

C) Calculate autocorrelation

17. What is the primary advantage of using an ensemble of models, such as Rando m Forest or Gradient
Boosting Machines?

A) Faster training time

B) Simpler model interpretation

C) Improved prediction accuracy

18. W hich technique is used for text data preprocessing in N LP , where words are converted to their base or
root f o r m?

A) Tokenization

B) Lemmatization

C) Bag-of-Words

19. Which regularization term is added to the cost function during training to penalize large weights in a
neural network?

A) L1 regularization (Lasso)

B) L2 regularization (Ridge)

C) ElasticNet regularization

20. Which algorithm is commonly used for finding the optimal number of clusters in a dataset when the
number of clusters is unknown?

A) K-means clustering

B) Hierarchical clustering

C) Silhouette analysis

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

21. In time series analysis, what is the autocorrelation of a time series at lag k?

A) Time series lagged correlation.

B) Cross-series lagged correlation.

C) Time series moving correlation.

22. W hich method is used for hyperparameter tuning b y exhaustively trying all possible combinations o f
h yp e r p ar a me t er va lu es?

A) Grid Search

B) Random Search

C) Bayesian Optimization

23. In a classification problem, which metric can be used to evaluate the model's ability to distinguish
b e tween positive a n d negative s a mp le s?

A) F1 score

B) Mean Absolute Error (MAE)

C) R-squared (R2) score

24. W hich technique is used to prevent data leakage w h e n scaling numerical features before feeding them
into a machine learning model?

A) Min-Max scaling

B) Standardization

C) Scaling after train-test split

25. In natural language processing, what does N E R stand for?

A) Named Entity Recognition

B) Natural Entity Retrieval

C) Normalized Entity Resolution

26. Which algorithm is commonly used for document clustering in unsupervised learning?

A) Decision Tree

B) K-means clustering

C ) Random Forest

26. Which algorithm is commonly used for document clustering in unsupervised learning?

A) Decision Tree

B) K-means clustering

C ) Random Forest

27. In a decision tree, which attribute selection measure aims to maximize the information gain?

A) Gini impurity

B) Entropy

C) Variance

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

28. W h ich method is used for reducing the nu mbe r of features in a dataset while preserving the most
important information?

A) Principal Component Analysis (PCA)

B) Forward Feature Selection

C) Recursive Feature Elimination (RFE)

29. In deep learning, which activation function is often used in the output layer of a binary classification
neural network?

A) Sigmoid

B) Tanh

C) ReLU

30. W hich machine learning algorithm is used for generating synthetic samples to balance class distribution
in a n imbalanced dataset?

A) Decision Tree
B) SMOTE (Synthetic Minority Over-sampling Technique)

C) AdaBoost

31. In Bayesian inference, w h a t term is used to represent the prior belief about a model parameter before
seeing the data?

A) Posterior

B) Likelihood

C) Prior

32. W h i c h algorithm is used to detect patterns o r associations b e twe en items in a transactional database?

A) K-means clustering

B) Apriori algorithm

C) Support Vector Machines (SVM)

33. In time series analysis, w h a t is the difference between seasonal and cyclic patterns?

A) Seasonal: Fixed period, cyclic: Variable period.

B) Seasonal: Short duration, cyclic: Longer duration.

C) Seasonal: More predictable, cyclic: Less predictable.

34. Which algorithm is an extension of linear regression that can handle multiple outputs, making it suitable
for multi-output regression tasks?

A) Logistic Regression

B) Lasso Regression

C) Multivariate Linear Regression

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

35. In unsupervised learning, what is the objective of clustering algorithms?

A) Predict a target variable based on input features.

B) Group similar data points together based o n their characteristics.

C) Minimize the error between predicted and actual values.

36. W hich regularization te r m adds a penalty based on the absolute value of the model coefficients in linear
regression?

A) L1 regularization (Lasso)

B) L2 regularization (Ridge)

C) ElasticNet regularization

37. In Natural Language Processing (NLP), what is the purpose of tokenization?

A) Preprocess text data by converting words to lowercase.

B) Split the text into individual words or tokens.

C) R e m o v e punctuation a nd special characters f r o m the text.

38. W hich ensemble learning method combines predictions from multiple models through a weighted voting
system to m ak e final predictions?

A) Bagging

B) Boosting

C) Stacking

39. In time series analysis, what is the order of differencing used to ma k e a non-stationary time series
stationary?

A) First-order differencing

B) Second-order differencing

C ) It depends o n the series characteristics

40. Which machine learning algorithm is suitable for solving multi-class classification problems b y reducing
them into multiple binary classification tasks?

A) K-nearest neighbors (KNN)

B) Logistic Regression

C) Decision Tree

41. In deep learning, what is the role of the learning rate in the training process?

A) Hidden layer control.

B) Output layer size.

C) Weight update step size.

42. W hich method is used to measure the similarity between two documents in natural language processing?

A) Euclidean distance

B) Jaccard similarity

C) Pearson correlation coefficient

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

43. In reinforcement learning, wha t is the term used for the agent's strategy for selecting actions to maximize
the cumulative rew ard over time?

A) Policy

B) Value Function

C) Q-function

44. Which algorithm is commonly used for regression tasks when the target variable exhibits a non-linear
relationship w ith the input features?

A) Linear Regression
B) Decision Tree

C) K-means clustering

45. In machine learning, what is the purpose of cross-validation?

A) Multi-dataset training.
B ) Inde pe ndent dataset assessment.

C) Train-test data division.

46. W hich method is used for dimensionality reduction in feature space b y transforming data to a lower-
dimensional representation that preserves pairwise distances?

A) t-SNE (t-distributed Stochastic Neighbor Embedding)

B) LDA (Linear Discriminant Analysis)

C) Isomap (Isometric Mapping)

47. In a classification problem, which metric is used to evaluate the model's ability to correctly classify
positive s a mp les?

A) Mean Squared Error (MSE)

B) Precision

C) R-squared (R2) score

48. W h i c h meth od is used for feature selection b y recursively r e mo v ing the least important features based
o n mo d e l performance?

A) Recursive Feature Elimination (RFE)

B) Principal Component Analysis (PCA)

C) Variance Thresholding

49. In natural language processing, wh a t is the term for reducing words to their base or root for m to capture
their core meaning?

A) Tokenization

B) Lemmatization

C) Stopword Removal

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

50. W hich algorithm is c o mmo n ly used for time series forecasting and can handle both trend and seasonality
in the data?

A) Support Vector Machines (SVM)

B) K-nearest neighbors (KNN)

C) Autoregressive Integrated Moving Average (ARIMA)

51. In ensemble learning, which method uses multiple w e a k learners sequentially, where each learner
corrects th e errors of its pr ede c es sor?

A) Bagging
B) Boosting

C) Stacking

52. W h ich technique is used to preprocess text data b y converting all words to lowercase to ensure
consistent co mp ar isons ?

A) Tokenization

B) Lemmatization

C) Lowercasing

53. In deep learning, what is the role of the loss function in training a neural network?

A) Regularize the model to prevent overfitting.

B) Measure the model's prediction error during training.

C) Control the learning rate for gradient updates.

54. Which machine learning algorithm is well-suited for outlier detection and can be used for one-class
classification?

A) Random Forest

B) k-Nearest Neighbors (k-NN)

C) Support Vector Machines (SVM)

55. In time series analysis, what is the term for patterns that repeat at regular intervals, such as daily, weekly,
or yearly patterns?

A) Seasonality

B) Trend

C) Cyclical pattern

56. W h ich meth od is used for hyperparameter tuning that employs probabilistic mode ls to model the
objective function and decide wh ich hyperparameters to try next?

A) Grid Search

B) Random Search

C) Bayesian Optimization

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

57. In a classification problem, which metric is used to evaluate the model's ability to correctly identify
negative samples?

A) Mean Absolute Error (MAE)

B) Recall

C) R-squared (R2) score

58. W hich preprocessing technique is used to scale numerical features to a specific range, such as [0, 1]?

A) Min-Max scaling

B) Standardization

C) Normalization

59. W hich technique is used to handle missing data in a dataset?

A) Feature scaling

B) Data imputation

C) Principal Component Analysis (PCA) ans wer

60. W hich ensemble learning method combines multiple models through averaging or voting to ma k e final
predictions?

A) Bagging

B) Boosting

C) Stacking

61. In a time series, what is the difference between lagged variables and moving averages?

A) Lagged variables represent past.

B) Lagged variables represent future.

C) Different concepts, not interchangeable.

62. W h ich technique is used to prevent data leakage during feature engineering by fitting transformations
based only o n the training data?

A) Min-Max scaling

B) Standardization

C) Scaling after train-test split

63. In deep learning, what is the primary function of a dropout layer?

A) Introduce noise into the data to prevent overfitting.

B) Regularize the model by randomly dropping some neurons during training.

C) Reduce the dimensionality o f the data.

64. W hich algorithm is co mmo n ly used for text classification tasks when the text data is represented a s word
frequencies or TF-IDF vectors?

A) Decision Tree

B) k-Nearest Neighbors (k-NN)

C) Naive Bayes

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

65. In Bayesian inference, what is the term used for updating the prior belief with n e w data to obtain the
posterior distribution?

A) Maximum Likelihood Estimation (MLE)

B) Maximum a Posteriori (MAP) estimation

C) Bayes' Theorem

66. Which algorithm is commonly used for finding the optimal number of clusters in a dataset w h e n the
number of clusters is unknown?

A) K-means clustering

B) Hierarchical clustering

C) Silhouette analysis

67. In time series analysis, what is the term for a model that combines autoregressive (AR), moving average
(MA), and differencing components?

A) Autoregressive Integrated Moving Average (ARIMA)

B) Exponential Smoothing (ETS)

C) Seasonal Autoregressive Integrated Moving-Average (SARIMA)

68. Which machine learning algorithm is used for predicting continuous numeric values, such as house prices
or stock prices?

A) Logistic Regression

B) Decision Tree

C) Linear Regression

69. In ensemble learning, what is the purpose of bagging techniques, such as R a n d o m Forest?

A) Ensemble model decision combining.

B) Model averaging variance reduction.

C) Sequential boosting weak learners.

70. W h ich meth od is used for data preprocessing t o handle categorical variables b y converting them into
numerical form?

A) One-Hot Encoding

B) Standardization

C) Feature Scaling

71. In deep learning, which activation function is commonly used in the hidden layers of a neural network to
introduce non-linearity?

A) Sigmoid

B) Tanh

C) ReLU (Rectified Linear Unit)

72. Which machine learning algorithm is commonly used for binary classification problems when the data is
linearly separable?

A) Decision Tree

B) Naive Bayes

C) Logistic Regression

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

73. In reinforcement learning, what is the term used for the process of exploring the environment a nd
collecting experiences to improve the agent's policy?

A) Q-learning

B) Policy Gradient

C) Exploration-Exploitation trade-off

74. Which method is used to handle missing data by filling in missing values based on the average of the non-
missing values in the same column?

A) K-nearest neighbors (KNN)

B) Mean substitution

C) Multiple Imputation by Chained Equations (MICE)

75. In Natural Language Processing (NLP), which technique is used to convert words into their base form by
removing suffixes or prefixes?

A) Tokenization

B) Lemmatization

C) Stemming

76. W hich evaluation metric is commonly used for regression tasks to assess the variance explained by the
model's predictions?

A) F1 score

B) Mean Absolute Error (MAE)

C) R-squared (R2) score

77. In deep learning, what is the purpose of the optimizer in the training process of a neural network?

A) Initialize the model weights randomly.

B) Reduce the learning rate as the training progresses.

C) Update the model weights to minimize the loss function.

78. Which machine learning algorithm is suitable for solving multi-class classification problems by
transforming th em into multiple binary classification tasks?

A) K-nearest neighbors (KNN)

B) Logistic Regression

C) Decision Tree

79. In ensemble learning, which technique is used to combine multiple models into one weighted model
based o n their p erfor ma n ce?

A) Bagging

B) Boosting

C) Stacking

80. W hich method is used for dimensionality reduction b y mapping data points into a lower-dimensional
space while preserving pairwise distances?

A) Principal Component Analysis (PCA)

B) Linear Discriminant Analysis (LDA)

C) t-distributed Stochastic Neighbor Embedding (t-SNE)

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

81. In deep learning, which activation function is commonly used in the output layer of a multi-class
classification neural network?

A) Sigmoid

B) Tanh

C) Softmax

82. Which algorithm is commonly used for document clustering in unsupervised learning?

A) Decision Tree

B) K-means clustering

C) Random Forest

83. In a decision tree, which attribute selection measure aims to minimize the Gini impurity? 83. In a decision
tree, which attribute selection measure aims to minimize the Gini impurity?

A) Gini impurity

B) Entropy

C) Variance

84. W h ich meth od is used for reducing the nu mb e r of features in a dataset b y selecting the most important
ones based o n statistical tests or scoring functions?

A) Principal Component Analysis (PCA)

B) Forward Feature Selection

C) Recursive Feature Elimination (RFE)

85. In deep learning, what is the role of the dropout technique during training a neural network?

A) Data noise for overfitting.

B) Dropout regularization for neurons.

C) Learning rate control gradient.

86. Which machine learning algorithm is used for predicting continuous numeric values, such as house prices
or stock prices?

A) Logistic Regression

B) Decision Tree

C) Linear Regression

87. In natural language processing, what is the term for reducing wo rds to their base or root form to capture
their core meaning?

A) Tokenization

B) Lemmatization

C) Stopword Removal

89. In a time series, what is the difference between lagged variables and moving averages?

A) Lagged: Past, Moving: Window.

B) Lagged: Future, Moving: Window.

C) Different concepts, not interchangeable.

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

90. W hich technique is used to prevent data leakage during feature engineering b y fitting transformations
based only o n the training data?

A) Min-Max scaling

B) Standardization

C) Scaling after train-test split

91. Which algorithm is a variant of the Rando m Forest algorithm and uses gradient boosting for improved
p e r f o r ma nce?

A) XGBoost

B) Naive Bayes

C) K-means

92. W hich algorithm is c o mmo n ly used for text classification tasks wh en the text data is represented as word
frequencies or TF-IDF vectors?

A) Decision Tree

B) k-Nearest Neighbors (k-NN)

C) Naive Bayes

93. In Bayesian inference, what is the term used for updating the prior belief with n e w data to obtain the
posterior distribution?

A) Maximum Likelihood Estimation (MLE)

B) Maximum a Posteriori (MAP) estimation

C) Bayes' Theorem

94. Which algorithm is commonly used for finding the optimal number of clusters in a dataset wh en the
number of clusters is unknown?

A) K-means clustering

B) Hierarchical clustering

C) Silhouette analysis

95. In time series analysis, what is the term for a model that combines autoregressive (AR), moving average
(MA), and differencing components?

A) Autoregressive Integrated Moving Average (ARIMA)

B) Exponential Smoothing (ETS)

C) Seasonal Autoregressive Integrated Moving-Average (SARIMA)

96. Which machine learning algorithm is commonly used for binary classification problems w h e n the data is
linearly separable?

A) Decision Tree

B) Naive Bayes

C) Logistic Regression

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

97. In reinforcement learning, wh a t is the term used for the process of exploring the environment a n d
collecting experiences to improve the agent's policy?

A) Q-learning

B) Policy Gradient

C) Exploration-Exploitation trade-off

98. Which method is used to handle missing data by filling in missing values based on the average of the non-
missing values in the same column?

A) K-nearest neighbors (KNN)

B) Mean substitution

C) Multiple Imputation by Chained Equations (MICE)

99. In Natural Language Processing (NLP), which technique is used to convert words into their base form by
removing suffixes or prefixes?

A) Tokenization

B) Lemmatization

C) Stemming

100. W hich evaluation metric is c o mmo n ly used for regression tasks to assess the variance explained b y the
model's predictions?

A) R-squared (R2) score

B) Mean Absolute Error (MAE)

C) F1 score

101. In deep learning, wh a t is the purpose of the optimizer in the training process of a neural network?

A) Initialize the model weights randomly.

B) Update the model weights to minimize the loss function.

C) Reduce the learning rate as the training progresses.

102. Which machine learning algorithm is suitable for solving multi-class classification problems by
transforming th em into multiple binary classification tasks?

A) K-nearest neighbors (KNN)

B) Logistic Regression

C) Decision Tree

103. In ensemble learning, which technique is used to combine multiple models into one weighted model
based o n their p erfor ma n ce?

A) Bagging

B) Boosting

C) Stacking

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

104. Which method is used for dimensionality reduction b y mapping data points into a lower-dimensional
space while preserving pairwise distances?

A) Principal Component Analysis (PCA)

B) Linear Discriminant Analysis (LDA)

C) t-distributed Stochastic Neighbor Embedding (t-SNE)

105. W hich method is used for handling missing data by creating multiple imputations based o n the
o b s e r v e d d at a?

A) K-nearest neighbors (KNN)

B) Mean substitution

C) Multiple Imputation by Chained Equations (MICE)

106. Which algorithm is commonly used for document clustering in unsupervised learning?

A) Decision Tree

B) K-means clustering

C) Random Forest

107. In a decision tree, which attribute selection measure aims to minimize the Gini impurity?

A) Gini impurity

B) Entropy

C) Variance

108. W h ich meth od is used for reducing the nu mb er of features in a dataset b y selecting the most important
ones based o n statistical tests or scoring functions?

A) Principal Component Analysis (PCA)

B) Forward Feature Selection

C) Recursive Feature Elimination (RFE)

109. In Reinforcement Learning, what is the n a m e of the technique used to estimate the expected long-term
reward from a given state?

A) Q-learning

B) Policy Gradient

C) Value Function

110. Which machine learning algorithm is used for predicting continuous numeric values, such as house
prices or stock prices?

A) Logistic Regression

B) Decision Tree

C) Linear Regression

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

111. W hat are autoencoders in deep learning?

Models used for reinforcement learning

N e tw o r k s designed to reconstruct inpu t data

Neural networks used for unsupervised learning

112. W hich layers are typically present in autoencoders?

Encoder and Classifier layers

Input an d Ou tpu t layers

Encoder and Decoder layers

113. T w o practical usages of autoencoders are:

Image denoising, Anomaly detection

Image classification, Speech recognition

Text generation, Language translation

114. What is an activation function in deep learning?

A function to convert nume r ic data to categorical data

A function used to define the output of a neuron

A function used to define the hidden layers

115. Three different types of activation functions are:

Linear, ReLU, and Sigmoid

Softmax, Sigmoid, and Step

Sigmoid, Tanh, and Leaky ReLU

116. To reduce overfitting in a deep neural network, you can:

Increase the n u mb e r of layers a nd neurons

U se dropout, regularization, a n d early stopping

Train the model for mo r e epochs without validation

117. W h y should w e use Batch Normalization in deep learning?

T o stabilize and accelerate the training process

T o increase the complexity of the model

T o reduce the training time of the model

118. If your model is suffering from the problem of Exploding Gradients, what might happen during training?

Loss and gradients b e c o me very large

Loss and gradients b ecome very small

The model converges quickly to the global min imu m

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

119. A few hyperparameters used for training a neural network are:

Activation function, Dropout, Regularization strength

Learning rate, Batch size, and Nu mb er of epochs

Momentum, Learning rate, and Hidden layer sizes

120. Parameter sharing in deep learning refers to:

Sharing parameters b e t w e e n input a n d output layers

Sharing parameters be tw een different types of layers

Sharing parameters b e tw e e n layers of the s a me type

121. The architecture of a typical Convolutional Neural Network (CNN) includes:

Convolutional, Pooling layers, Fully connected layers

Only fully connected layers

Only pooling layers

122. What is the Vanishing Gradient Problem in Artificial Neural Networks?

W h e n gradients in backpropagation be co me too small

W h e n gradients in backpropagation b e c o me too large

W h e n the learning rate becomes very small

123. If the loss doesn't decrease in a few epochs while training a n artificial neural network, it could be due to:

A n issue with the data preprocessing or normalization

A large learning rate causing divergence

Overfitting o n the training data

124. W h y Sigmoid or Tanh is not preferred in the hidden layer of a neural network?

They are computationally expensive

Th e y are not differentiable

They cause the vanishing gradient problem

125. Transfer learning is recommended when:

T h e source and target ha ve different input do ma ins

T h e source an d target tasks are unrelated

T h e target task has a large labeled dataset

126. What is the vanishing gradient problem in RNN, and how can it be solved?

The gradients b eco me too large during training

Th e mod el is unable to learn n e w patterns

The gradients b eco me too small during training

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

127. The main gates in LSTM (Long Short-Term Memory) are:

Input gate, Output gate, and Forget gate

Input gate, Output gate, a n d Upd a te gate

Input gate, Forget gate, and Update gate

128. Using a CNN to classify 1D signals:

Is not recommended, a s C N N s are designed for image

Is suitable only for binary classification tasks

Is effective if the 1 D signals represent sequential data

129. L1/L2 regularization affects a neural network by:

Penalizing large weights to avoid overfitting

Reducing the model size

Increasing the learning rate during training

130. H o w to change a pre-trained neural network from classification to regression?

Chang e the loss function

Replace the activation function with a linear activation

A d d mor e hidden layers to the network

131. Setting the mo me n tu m hyperparameter too close to 1 (e.g., 0.9999) wh en using an S G D optimizer can
lead to:

Overshooting the global min i mu m and oscillations

Faster convergence of the mo d e l

Slower learning and longer training times

132. Hyperparameters that can be optimized for the batch normalization layer are:

Learning rate and Dropout rate

Scale factor an d Center factor

Activation function and Optimizer

133. Dropout affects the training and prediction speed of a deep learning mo d e l by:

Decreasing training a nd increasing prediction speed

Increasing training and decreasing prediction speed

Having no impact o n either

134. Th e advantage of deep learning over traditional machine learning is:

Th e ability to learn hierarchical features f ro m data

Simplicity of model architecture

L o w er computational requirements

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

135. A depthwise Separable layer in a CNN:

Is used for global feature extraction

Co mb in es depthwise a n d pointwise convolutions

Performs pooling operations for downsampling

136. Siamese Network is beneficial in addressing the one-shot learning problem because it:

Uses multiple G P U s for parallel processing

Utilizes transfer learning to improve model

Encodes similarity between inputs like shot learning

137. Grayscaling serves the purpose of:

Converting colored images to grayscale

Reducing image dimensionality

Enhancing the contrast of images

138. Translational equivariance me a n s that a model:

P roduces s a m e output for translated versions o f input

C a n translate inputs to different languages

Is invariant to translation invariance

139. YO LO (You Only Look Once) is an object detection algorithm that:

Divide the image into grid and predicts bounding boxes

Processes the ima g e in multiple passes

Uses a sliding w indow to detect objects in the image

140. Dropouts in neural networks are used for:

Re mo v ing irrelevant features fro m the input data

Preventing overfitting b y deactivating neurons

Reducing the learning rate during training

141. Exploding gradients occur when:

The model fails to learn the training data

T h e mod el underfits the training data

Th e gradients during backpropagation b e c o m e large

142. According to you, which one is more powerful — a two-layer neural network without any activation
function or a two-layer decision tree?

The two-layer neural network

Both have similar power, but depends on the problem

The two-layer decision tree

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

143. W hile building a neural network architecture, h o w will y ou decide h o w m a n y neurons a n d hidden layers
the neural n e t w o r k sh o u l d h a v e?

A d d a s m a n y neurons a n d hidden layers as possible

U s e a trial-and-error m e t h o w i t h various architecture

C h o o s e the neur on s b a s e d o n size of th e training dat a

144. In the context of deep learning, wh a t is a tensor?

A type of activation function

A vector representation of the weights in a N N

A multi-dimensional array used to store data

145. W h at are the advantages of using transfer learning in deep learning?

Faster training an d better generalization

I mp roved interpretability o f the mo d e l

L o w er computational requirements

146. What is the role of multi-class classification problems?

Multi-class involves mo re than two classes Multi-

class assigns multiple labels to a single instance N o n e

of the above

147. W h at are the different techniques to achieve data normalizationial?

Min-Max scaling, Standardization, and L1 normalization

Feature scaling and L2 normalization

Z-score normalization and L1 regularization

148. W hat is forward propagation in the context of deep learning?

Fo rward propagation updates the weights,

Forward propagation gives the gradients during training

N o n e of the above

149. N a m e the different types of deep neural networks.

Support Vector Machines and Logistic Regression K-

Means Clustering and Principal Component Analysis

Feedforward Neural Networks, RNN, CNN and GAN

150. What is the benefit of using Re LU activation over Sigmoid or Tanh?

R e L U has better numerical stability

R e L U produces a continuous output in the range [0, ∞ ]

R e L U doesn't suffer from vanishing gradient problem

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

151. Explain the concept of depthwise Separable layers and their advantages in CNNs.

They combine depthwise and pointwise convolutions

Depthwise Separable layers apply pooling operations

Depthwise Separable layers utilize weight sharing

152. H o w can Siamese Network be beneficial in addressing the one-shot learning problem?

It uses multiple G P U s for parallel processing

It utilizes transfer learning to improve performance

Siamese Network encodes similarity between inputs

153. Grayscaling an image serves the purpose of:

Converting colored images to grayscale

Reducing image dimensionality

Enhancing the contrast of images

154. Explain the concept of translational equivariance in deep learning

The model can change inputs to different languages

It is the same as invariance to translation invariance

It refers to the ability to recognize translated versions

155. YO LO (You Only Look Once) is an object detection algorithm that:

Processes the image in multiple passes

Divides the image into a grid

Uses a sliding window to detect objects in the image

156. What do you know about dropouts in neural networks?

Dropouts are used for removing irrelevant features

Dropouts reduce the learning rate during training

Dropouts prevent overfitting b y deactivating neuron s

157. Explain the concept of exploding gradients in deep learning.

Occur w h en gradients b e c o me too small

Occur w h en gradients in backpropagation b eco me large

S a me as vanishing gradient

158. What is Bias Measure?

Bias measures the ability to generalize to n e w data

Bias measures the model's sensitivity to training data

Bias and variance are the s a me concepts

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

159. Which one is more powerful — a two-layer neural network without any activation function or a two-layer
decision tree?

The two-layer neural network

The two-layer decision tree

Both have similar power, but it depends o n problem

160. While building a neural network architecture, h o w will you decide h o w many neurons and hidden layers
the neural network should have?

Ch o o s e the n u mb e r of neurons b as ed o n size of dataset

A d d as ma n y neurons and hidden layers as possible

U s e a trial-and-error appro ach

161. In the context of deep learning, what is a tensor?

A multi-dimensional array used to store data

A type of activation function

A vector representation of the weights in a N N

162. W h at are the advantages of using transfer learning in deep

learning?

I mp roved interpretability o f the mo d e l

Faster training an d better generalization

L o w er computational requirements

163. What is NLG in NLP?

Natural Language Generation

Natural Language Graph

Natural Language Generator

164. W h at is the order of steps in natural language understanding?

Syntactic analysis, Signal analysis, Pragmatic analysis

Signal processing, Syntactic analysis, Semantic analysis

Pragmatic analysis, Signal analysis, Syntactic analysis

165. What is the purpose of signal processing in NLP?

T o preprocess raw text data for analysis

To analyze sentiment in text

T o convert audio signals to text

166. What does pragmatic analysis in NL P involve?

Analyzing the context and intention of the speaker

Extracting entities from text

Identifying the grammatical structure of sentences

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

167. What is the main focus of syntactic analysis in NLP?

Extracting n a m e d entities f r o m text function

Analyzing the g r a m ma r a nd structure of sentences

Understanding the me a ni n g of sentences

168. What is the goal of semantic analysis in NLP?

Determining the sentiment of a text

Identifying the subject and object in a sentence

Understanding the me a n i n g and intent of sentences

169. What does sentiment analysis in N L P involve?

Analyzing the emotions expressed in a text

Detecting the grammatic al errors in a sentence

Identifying the main topics in a text

170. What is discourse analysis in NLP?

U n d er s t an d in g th e structure a n d f low o f a conve rs a tion

Identifying coreference b e t w e e n p r o n o u n s & entities

An alyzing relationships b e tw e e n w o r d s in a sentence

171. What is pragmatic ambiguity in NLP?

Ambiguity related to context a nd speaker's intention

Ambiguity caused by multiple meanings of a word

Ambiguity du e to errors in the text corpus

172. Which of the following is a major application of NLP?

I ma g e recognition

M a c h i n e translation

S p e e c h synthesis

173. Real-world applications of NLP include:

Object detection in images

W eath er forecasting

Sentiment analysis of customer reviews

174. What are the c o mmo n N L P techniques used for?

I ma g e manipulation

Data visualization

Language understanding a n d text analysis

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

176. W hat are the main components of N L P ?

Semantic analysis a nd Discourse analysis

Natural Language Understanding & N L G eneration

Signal processing and Syntactic analysis

177. Which tools are commonly used for training N L P models?

TensorFlow and PyTorch

NumPy and SciPy

NLTK and spaCy

178. Which N L P technique uses a lexical knowledge base to obtain the

correct base for m of word s?

W ordNet-b as ed normalization

Lemmatization

Stemming

179. Which models are commonly used to reduce the dimensionality of data in NLP?

S V M and Random Forest

LSA & LDA

Naive Bayes and Decision Trees

180. What does a masked language model do in NLP?

Fills in missing words in a sentence

R e mo v e s stop word s fro m the text

Replaces ambiguous words with appropriate mea ning

181. Which of the following is an open-source library for NLP?

Microsoft Word

Natural Language Toolkit (NLTK)

TensorFlow

182. What is the purpose of the bag of words model in NLP?

T o represent data as frequency distribution o f w ords

T o represent the text data as a collection o f wo rds

T o count the occurrence of each w o r d in the text

183. What does CBOW stand for in NLP?

Continuous Bag of Words

Contextual Bag of Words

Comprehensive Bag of Words

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

184. What is TF-IDF used for in NLP?

T o r epr esent text d a t a a s v ect o r s

T o w e i g h the importance of w o r d s in a d o cu ment

T o me a s u r e the similarity b e t w e e n d o cu me n t s

185. What do POS and tagging stand for in NLP?

Position of Sentence and Text Analysis Grouping

Part of Speech and Text Aggregation

Part of Speech and Tagging

186. What is an n-gram in NLP?

A combination of n words used for text analysis

A type of language model

A sequence of n words in a sentence

187. What is skip-gram in NLP?

A word2vec model for predicting context words

A technique to skip irrelevant words in a sentence

A model used for skip-gram language learning

188. In NLP, what is a corpus?

A collection of raw text documents used for analysis

A statistical measure of word frequency

A technique to convert sp ee ch to text

189. What are the features of the text corpus in N L P?

File format, word length, a n d character encoding

W o rd tokens and paragraph d ivisions

P O S tags, word frequency, and word embeddings

190. What is normalization used for in NLP?

T o bring data to a c o m m o n format

T o standardize text documents for comparison

T o process speech data for analysis

191. What is keyword normalization in NLP?

Converting keywords to lowercase for consistency

Transforming keywords to their base forms

Removing non-relevant words (stop words)

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

192. What is lemmatization in NLP?

T h e process of replacing rare w o r d s w ith c o m m o n

T h e process of reducing a w o r d to its base or root f o r m

T h e process of converting text into n u mer i cs

193. What is stemming in NLP?

The process of finding the most relevant keywords

T h e process of parsing sentences into g r a mma r

The process of grouping words with the s a me root

194. What does Open CV stand for?

Open Computer Vision

Open Code Vision

Optical Character Verification

195. What is OpenCV primarily used for?

Image and video processing

Audio processing

Natural Language Processing

196. Which programming languages are officially supported by OpenCV?

Python and C++

Java and R

C # and JavaScript

197. What is the role of OpenCV's "cv2" module in Python?

Handling file I/O operations

Image and video processing

Mathematical computations

198. Which of the following data types is commonly used to represent images in Op en CV?

Array of integers

Dictionary of pixels

N u mp y array

199. What is the purpose of image thresholding in OpenCV?

Converting a n image to grayscale

Applying a color map to the image

Converting an image to binary

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

200. W h ich O p e n C V function is used to detect edges in an image?

cv2.erode()

cv2.Canny()

cv2.dilate()

201. What is the "kernel" in OpenCV's image processing functions?

A part of an image

A function for color mapping

A small matrix used for convolution

202. What is histogram equalization used for in image processing?

Enhancing the contrast of an image

Removing noise from an image

Applying filters to an image

203. Which Op enCV function is used to apply a Gaussian blur to an image?

cv2.MedianBlur()

cv2.GaussianBlur()

cv2.boxFilter()

204. How does the Hough Line Transform work in OpenCV?

It detects straight lines in an image

It detects edges in an image

It detects circles in an image

205. Which Op enCV function is used to perform image resizing?

cv2.scale()

cv2.resize()

cv2.transform()

206. What is the role of the cv2.VideoCapture() function in OpenCV?

Capturing video from a camera or a file

Capturing images from a w ebc a m

Recording audio from a microphone

207. W hat is the purpose of feature detection algorithms in O p e n CV?

Detecting objects or keypoints in an image

Enhancing the color balance of an image

Removing noise from an image

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

208. W h ich O p e n C V function is used for drawing shapes and text on an image?

cv2.drawContours()

cv2.rectangle()

cv2.putText()

209. What is the role of the cv2.matchTemplate() function in OpenCV?

Finding the similarity between two images

Calculating the histogram of an image

Template matching to locate objects in an image

210. H o w does the cv2.findContours() function work in OpenCV?

It finds and segments regions with similar intensity

It finds the edges of an image

It finds the mi n i mu m area bounding rectangle

211. What is the purpose of the cv2.CascadeClassifier() in OpenCV?

It classifies images into predefined categories

It detects faces or objects in an image or video

It performs morphological operations on images

212. What is the role of the cv2.goodFeaturesToTrack() function in OpenCV?

Detecting corners in an image

Tracking objects in a video stream

Calculating the optical flow between two frames

213. Which Op enCV function is used for image bitwise operations?

cv2.bitwise_and()

cv2.bitwise_xor()

cv2.bitwise_or()

214. What is the purpose of the cv2.SIFT() function in OpenCV?

Detecting and extracting SIFT keypoints from an image

Performing image rotation and scaling

Applying a Sobel filter to an image

215. W hich O p e n C V function is used to compute the perspective transformation of an image?

cv2.getPerspectiveTransform()

cv2.getAffineTransform()

cv2.getPerspective()

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

216. What is the role of the cv2.warpAffine() function in OpenCV?

Applying an affine transformation to an i mag e

Resizing an image

Converting a n i mag e to grayscale

217. What does the cv2.fastNlMeansDenoising() function do in OpenCV?

Enhances the sharpness of an ima ge

Adjusts the brightness a n d contrast of an ima g e

Removes noise from an image

218. H o w does the cv2.SimpleBlobDetector() function work in OpenCV?

It detects a n d extracts b lo b s or c o n ne c te d c o mp o n e n t s

It detects simple shapes like circles a nd squares

It computes the center of ma s s of an object in an image

219. Which O p e n C V function is used to apply a perspective transform to an image?

cv2.warpPerspective()

cv2.warpAffine()

cv2.perspectiveTransfor m()

220. What is the purpose of the cv2.inRange() function in OpenCV?

Creating a binary ma s k based on pixel intensity range

Finding the range of pixel intensities in an image

Detecting regions of interest in an image

221. H o w does the cv2.SimpleBlobDetector_Params() object help in OpenCV?

It stores parameters for ima g e rotation a n d scaling

It stores parameters for the blob detection algorithm

It stores parameters for ima g e s mooth ing

222. What is the purpose of cv2.getPerspectiveTransform() in OpenCV?

T o find the perspective transformation matrix

To perform a n affine transformation on an image

For perspective transformation matrix in image warping

223. H o w does the cv2.createBackgroundSubtractorMOG2() function work in OpenCV?

It creates a background m a s k for object detection

It creates a 3 D model of the image foreground

It creates a Gaussian Mixture-based background

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

224. W hat is the role of the cv2.remap() function in Op en CV?

It r e map s a n i ma g e to a different color spa c e

It r e map s a n i m a g e f r o m o n e coordinate to another

It re map s an imag e using bilinear interpolation

225. Which Op enCV function is used to compute the optical flow of an image?

cv2.calcOpticalFlowFarneback()

cv2.calcOpticalFlowSingleScale()

cv2.calcOpticalFlowPyrLK()

226. What is the purpose of the cv2.HoughCircles() function in OpenCV

Detecting circles in an image

Detecting lines in an image

Detecting corners in an image

227. How does the cv2.KAZE() feature detector work in OpenCV?

It detects and extracts keypoints f r o m an ima g e

It detects affine invariant keypoints in an image

It detects edges in an image

228. What is the purpose of the cv2.matchShapes() function in Op enCV?

It measures the similarity be tw e en tw o shapes

It matches similar shapes in two images

It finds the contour similarity between two images

229. What is the primary purpose of the A d a m optimizer in deep learning?

Regularization of the neural network

Adaptive learning rate optimization

Batch normalization of the input data

230. W h i c h type of neural network architecture is best suited for sequential data and natural language
processing tasks?

Recurrent Neural Network (RNN)

Feedforward Neural Network

Convolutional Neural Network

231. W hich technique is used to avoid the problem of exploding gradients in deep neural networks?

Gradient clipping

L1 regularization

Dropout regularization

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.
 https://pianalytix.com/ https://instagram.com/pianalytix https://www.linkedin.com/company/pianalytix/

232. In a long short-term me mo r y ( L S T M ) cell, which gate is responsible for deciding what information to
forget f r o m t he pr ev i o u s t i me step?

Input gate

Forget gate

Output gate

233. In deep reinforcement learning, what is the function of the "Q-network"?

It calculates th e gradients for updating the mo d e l

It determines the exploration rate during training

It m a p s states to actions a n d predicts future rewa rds

234. W hat is the "batch size" in stochastic gradient descent (SGD) optimization for neural networks?

N u mb e r of training samplesbefore e ac h weight update

The number of epochs during training

Th e learning rate used for weight updates

235. W hich neural network architecture is designed to handle variable-length input sequences b y applying
the s a me weights at every time step?

Recurrent Neural Network

Feedforward Neural Network

Convolutional Neural Network

236. In deep reinforcement learning, what is the "policy network" responsible for?

It m a p s states to actions, specifying the agent's actions

It calculates discounted future rew ard s for e a ch action

It determines the exploration rate during training

237. The technique of "weight initialization" in neural networks is crucial for:

Avoiding overfitting during training

Ensuring convergence during training

Preventing the vanishing gradient problem

238. W hat is the primary advantage of using batch normalization in deep neural networks?

It reduces the computation time during training

It increases th e model's capacity

It improves the generalization of the mo d e l to n e w data

239. W h ich technique is used to reduce the dimensionality o f the feature space in neural networks b y
projecting it into a lower-dimensional subspace?

Batch Normalization

Gradient Boosting

Principal Component Analysis (PCA)

© 2023 All rights reserved. Pianalytix Edutech Pvt. Ltd. retains the copyright for this material worldwide.