Questions

Question 1: What is the primary goal of Data Science?

 A. To manually collect data from various sources
 B. To build machines that mimic human intelligence
 C. To analyze data for insights and solve complex problems (Answer)
 D. To predict outcomes without using data
Question 2: Which of the following is a characteristic of Artificial
Intelligence (AI)?
 A. AI systems can learn and adapt based on data (Answer)
 B. AI systems always require labelled data to function
 C. AI systems only work in the field of robotics
 D. AI systems replace human intelligence entirely
Question 3: In Machine Learning (ML), what does the term "training a
model" refer to?
 A. Feeding a model real-time data to make decisions
 B. Optimizing the model parameters to minimize error (Answer)
 C. Using a pre-built system for predictions
 D. Visualizing data for exploration
Question 4: How does Data Science help in healthcare?
 A. By improving website designs
 B. By analyzing patient data for personalized treatment (Answer)
 C. By creating video game simulations
 D. By automating mechanical engineering processes
Question 5: Which industry uses Data Science for fraud detection?
 A. Entertainment
 B. Retail
 C. Finance (Answer)
 D. Education
Question 6: What is one of the key applications of Data Science in
marketing?
 A. Generating random sales data
  B. Building customer segmentation models for targeted campaigns
(Answer)
 C. Automating stock prices
 D. Reducing expenses for raw materials
Question 7: Which application of Data Science benefits social media
platforms?
 A. Recommending personalized content to users (Answer)
 B. Designing augmented reality hardware
 C. Automating language translations
 D. Generating robotic assistants
Question 8: How is Data Science applied in sports analytics?
 A. To design stadium architecture
 B. To analyze player performance and optimize team strategies (Answer)
 C. To track ticket sales
 D. To create player uniforms
Question 9: Which role in Data Science focuses on cleaning, organizing,
and analyzing data to extract insights?
 A. Data Engineer
 B. Data Scientist (Answer)
 C. Machine Learning Engineer
 D. AI Researcher
Question 10: What is the primary responsibility of a Machine Learning
Engineer?
 A. Building and deploying machine learning models (Answer)
 B. Designing user interfaces for AI systems
 C. Managing databases
 D. Writing research papers
Question 11: Which skill is essential for a Data Scientist to manipulate
and analyze large datasets?
 A. Knowledge of cloud computing
 B. Proficiency in Python or R (Answer)
 C. Experience with graphic design
 D. Expertise in hardware engineering
Question 12: What is the role of an AI Researcher?
  A. Developing new algorithms and models for AI systems (Answer)
 B. Managing data pipelines
 C. Creating marketing strategies
 D. Writing SQL queries
Question 13: Which of the following is a key skill for a Data Engineer?
 A. Building and maintaining data pipelines (Answer)
 B. Designing neural networks
 C. Writing creative content
 D. Conducting statistical analysis
Question 14: What is the primary focus of a Business Analyst in the field
of Data Science?
 A. Building machine learning models
 B. Translating data insights into business strategies (Answer)
 C. Managing cloud infrastructure
 D. Designing AI algorithms
Question 15: What is a critical skill for an AI Engineer?
 A. Proficiency in deep learning frameworks like TensorFlow or PyTorch
(Answer)
 B. Expertise in video editing
 C. Knowledge of mechanical systems
 D. Experience with social media marketing
Question 16: Which role in Data Science involves creating dashboards
and visualizations to communicate insights?
 A. Data Analyst (Answer)
 B. Machine Learning Engineer
 C. AI Researcher
 D. Software Developer

Question 17: Which of the following is NOT a type of Machine Learning?

 A. Supervised Learning
 B. Reinforcement Learning
 C. Semi-Supervised Learning
 D. Incremental Learning (Answer)
Question 18: In supervised learning, what does the model require?
 A. Unlabeled data
 B. Feedback for every action
 C. Labeled input and output data (Answer)
 D. Randomized predictions
Question 19: What is the key purpose of unsupervised learning?
 A. To map inputs to outputs directly
 B. To discover hidden patterns or structures in unlabeled data (Answer)
 C. To maximize rewards for actions
 D. To deploy predictive models
Question 20: Which of the following is an example of supervised
learning?
 A. K-Means Clustering
 B. Decision Trees for classification tasks (Answer)
 C. Dimensionality reduction using PCA
 D. Identifying customer segments
Question 21: In reinforcement learning, what guides the agent’s
learning?
 A. Loss function optimization
 B. A set of labeled examples
 C. Rewards and penalties from the environment (Answer)
 D. Minimizing variance in predictions
Question 22: Which step of the ML model lifecycle involves dividing the
data into training and testing datasets?
 A. Feature Engineering
 B. Model Deployment
 C. Data Preprocessing and Exploration (Answer)
 D. Hyperparameter Optimization
Question 23: What is the main difference between supervised and
unsupervised learning?
 A. Supervised learning works with labeled data, while unsupervised
learning works with unlabeled data (Answer)
 B. Unsupervised learning is computationally simpler
 C. Supervised learning cannot be used for prediction tasks
  D. Unsupervised learning works only with structured data
Question 24: In the ML model lifecycle, what is the purpose of model
evaluation?
 A. To collect additional data
 B. To assess the model’s performance on unseen data (Answer)
 C. To retrain the model periodically
 D. To normalize features
Question 25: Which task would most likely require reinforcement
learning?
 A. Detecting spam emails
 B. Training a chatbot to answer questions
 C. Teaching a robot to walk through trial-and-error (Answer)
 D. Predicting house prices
Question 26: Which Python library is most commonly used for
implementing machine learning algorithms?
 A. TensorFlow
 B. Pandas
 C. scikit-learn (Answer)
 D. OpenCV
Question 27: When splitting the dataset in Python, which function is
used to divide data into training and test sets?
 A. split_data()
 B. train_test_split() (Answer)
 C. data_split()
 D. set_divider()
Question 28: How do you fit a logistic regression model in scikit-learn?
 A. Use the LogisticRegression class and the fit() function (Answer)
 B. Use the DecisionTreeClassifier and set mode to "logistic"
 C. Use the LinearRegression class and set the solver to "sigmoid"
 D. Use the KMeans function with 2 clusters
Question 29: What is a commonly used evaluation metric for logistic
regression models?
 A. Mean Squared Error (MSE)
 B. R-squared value
  C. Accuracy, Precision, Recall, and F1-Score (Answer)
 D. Silhouette Score
Question 30: When fitting a logistic regression model, what does the
random_state parameter do?
 A. Optimizes the learning rate
 B. Ensures reproducibility of results by controlling random number
generation (Answer)
 C. Automatically scales features
 D. Tunes hyperparameters
Question 31. What is the primary building block of a neural network?
 A. Dataframe
 B. Node or Neuron (Answer)
 C. Kernel
 D. Epoch
Question 32. What is the purpose of the activation function in a neural
network?
 A. To normalize input data
 B. To control overfitting
 C. To introduce non-linearity in the model (Answer)
 D. To aggregate feature maps
Question 33. Which of the following is a type of neural network used
primarily for image recognition?
 A. Recurrent Neural Network (RNN)
 B. Convolutional Neural Network (CNN) (Answer)
 C. Feedforward Neural Network
 D. Transformer
Question 34. What is a key characteristic of Recurrent Neural Networks
(RNNs)?
 A. They have fully connected layers only
 B. They use convolutional layers for feature extraction
 C. They retain memory of previous inputs for sequential data processing
(Answer)
 D. They cannot process time-series data
Question 35. Which type of neural network is most suitable for tasks
like language modeling and machine translation?
 A. CNN
 B. Feedforward Neural Network
 C. Transformer (Answer)
 D. Restricted Boltzmann Machine
Question 36. What is a primary advantage of neural networks over
traditional machine learning algorithms?
 A. Reduced training data requirement
 B. Ability to learn complex, non-linear relationships automatically
(Answer)
 C. Lower computational cost
 D. Direct interpretability of model decisions
Question 37. Which application is neural networks NOT widely used for?
 A. Image recognition
 B. Speech synthesis
 C. Sorting algorithms (Answer)
 D. Natural language processing
Question 38. What is a major weakness of neural networks?
 A. High dependency on feature engineering
 B. Difficulty in handling unstructured data
 C. Black-box nature and lack of interpretability (Answer)
 D. Inability to scale for larger datasets
Question 39. What is a unique advantage of Convolutional Neural
Networks (CNNs) in image processing?
 A. They reduce feature map dimensionality through recurrence
 B. They automatically learn spatial hierarchies of features (Answer)
 C. They require sequential inputs
 D. They are only suitable for time-series data
Question 40. Which challenge is addressed by dropout in neural
networks?
 A. Underfitting
 B. Overfitting (Answer)
 C. Data drift
  D. Decreased training time

Question 41: What is the function of the Conv2D class in TensorFlow

when building a CNN?
 A. Applies pooling operations to the feature maps
 B. Creates convolutional layers for processing image data (Answer)
 C. Connects layers fully
 D. Reduces overfitting using dropout
Question 42: Which pooling method is most commonly used in CNNs to
downsample feature maps?
 A. Average Pooling
 B. Max Pooling (Answer)
 C. Global Pooling
 D. Feature Pooling
Question 43: In TensorFlow, which function is used to compile a CNN
model before training?
 A. [Link]()
 B. [Link]()
 C. [Link]() (Answer)
 D. [Link]()
Question 44: In TensorFlow, what does the parameter filters in the
Conv2D layer specify?
 A. The type of pooling operation
 B. The size of the input image
 C. The number of feature detectors or kernels (Answer)
 D. The stride of the convolution
Question 45: When training a CNN using TensorFlow, which class is used
for data augmentation?
 A. ImageDataGenerator (Answer)
 B. Dataset.from_tensor_slices
 C. Conv2D
 D. Sequential
Question 46. Which architecture forms the basis for most Large
Language Models (LLMs)?
  A. Recurrent Neural Networks (RNNs)
 B. Transformers (Answer)
 C. Convolutional Neural Networks (CNNs)
 D. Autoencoders
Question 47. Which of the following is an example of a Large Language
Model?
 A. T5
 B. GPT-3
 C. BERT
 D. All of the above (Answer)

Question 48. What is a major advantage of LLMs compared to

traditional NLP models?
 A. They require less computational power
 B. They can generate human-like, context-aware text (Answer)
 C. They do not rely on pretraining
 D. They need fewer parameters to function
Question 49. How are LLMs fine-tuned for specific tasks?
 A. By using completely new training datasets
 B. By re-training the entire model from scratch
 C. By adjusting weights on task-specific labeled data (Answer)
 D. By only using gradient descent for optimization
Question 50. What is a limitation of Large Language Models?
 A. Inability to handle structured data
 B. Computational and resource costs during training and deployment
(Answer)
 C. Lack of understanding for text-based tasks
 D. Small-scale applications

Additional Questions
---

#### **1. What is the primary purpose of Large Language Models (LLMs)?**
- A. To execute mathematical operations
- B. To process and generate human-like text **(Answer)**
- C. To encode images into numerical data
- D. To replace traditional databases

---

#### **2. Which mechanism is central to the functionality of Transformer

models?**
- A. Pooling
- B. Attention **(Answer)**
- C. Backpropagation
- D. Regularization

---

#### **3. Which pre-trained LLM is optimized for bidirectional context

understanding?**
- A. GPT
- B. Bloom
- C. BERT **(Answer)**
- D. T5

---

#### **4. Which activation function is commonly used in convolutional layers of

CNNs?**
- A. Tanh
- B. Sigmoid
- C. ReLU **(Answer)**
- D. Softmax

---

#### **5. In logistic regression, what is the range of the sigmoid function's
output?**
- A. [-1, 1]
- B. [0, ∞]
- C. [0, 1] **(Answer)**
- D. [-∞, ∞]

---

#### **6. What is the primary input to convolutional layers in a CNN?**

- A. Text embeddings
- B. Images or feature grids **(Answer)**
- C. Sequential data
- D. Categorical labels

---

#### **7. Which dataset in sklearn is commonly used for binary classification
problems?**
- A. Boston Housing
- B. Iris Dataset **(Answer)**
- C. California Housing
- D. 20 Newsgroups

---

#### **8. What does pooling in CNNs help achieve?**

- A. Reduces the spatial dimensions of feature maps **(Answer)**
- B. Adds non-linearity
- C. Increases the size of data
- D. Enhances resolution

---

#### **9. In logistic regression, which metric evaluates the trade-off between
precision and recall?**
- A. Accuracy
- B. ROC Curve
- C. F1-Score **(Answer)**
- D. Confusion Matrix

---

#### **10. Which optimization algorithm is commonly used to minimize the

loss function in neural networks?**
- A. Stochastic Gradient Descent (SGD) **(Answer)**
- B. Max Pooling
- C. Random Forest
- D. Naive Bayes

---

#### **11. What type of architecture does the GPT series use?**
- A. Encoder-only
- B. Decoder-only **(Answer)**
- C. Encoder-Decoder
- D. Recurrent Neural Network

---

#### **12. Which of the following is a characteristic of supervised machine

learning?**
- A. Works without labeled data
- B. Uses feedback in training **(Answer)**
- C. Does not require a training set
- D. Randomly predicts outputs

---

#### **13. Which function transforms a neural network's output into

probabilities in logistic regression?**
- A. ReLU
- B. Tanh
- C. Sigmoid **(Answer)**
- D. Softmax

---

#### **14. Which loss function is used in logistic regression?**

- A. Mean Squared Error (MSE)
- B. Cross-Entropy Loss **(Answer)**
- C. Hinge Loss
- D. Binary Cross Validation

---

#### **15. What is the primary role of filters in CNNs?**

- A. To optimize weights
- B. To detect features **(Answer)**
- C. To reduce dimensions
- D. To add activation

---
#### **16. What does feature engineering involve?**
- A. Automating testing of models
- B. Selecting, creating, or transforming features **(Answer)**
- C. Deploying models
- D. Training using batch sizes

---

#### **17. What is the main challenge in unsupervised learning?**

- A. Limited dataset size
- B. Lack of labeled data for evaluation **(Answer)**
- C. Complexity of models
- D. Overfitting

---

#### **18. Which model architecture is commonly used for sequential data like
text or speech?**
- A. CNN
- B. RNN **(Answer)**
- C. LSTM
- D. Autoencoder

---

#### **19. Which neural network is primarily used for processing images?**
- A. RNN
- B. CNN **(Answer)**
- C. LSTM
- D. Transformer

---
#### **20. In the workflow for ML projects, which step involves cleaning data?
**
- A. Data Collection
- B. Data Exploration and Preprocessing **(Answer)**
- C. Model Deployment
- D. Feature Engineering

---

#### **21. What does backpropagation aim to minimize?**

- A. Model accuracy
- B. Loss function **(Answer)**
- C. Training time
- D. Activation variance

---

#### **22. Which model popularized the concept of multi-head attention?**

- A. CNN
- B. Transformer **(Answer)**
- C. RNN
- D. Logistic Regression

---

#### **23. In supervised learning, what is the primary requirement for training
data?**
- A. Uniform distribution
- B. Labeled data **(Answer)**
- C. Feature dimensionality
- D. Reduced variance
---

#### **24. Which algorithm is typically used for binary classification in logistic
regression?**
- A. k-Means Clustering
- B. Decision Trees
- C. Logistic Regression **(Answer)**
- D. Neural Networks

---

#### **25. Which LLM is open-source and focuses on multilingual tasks?**

- A. GPT-3
- B. BERT
- C. Bloom **(Answer)**
- D. PaLM

---

#### **26. What is the main advantage of pooling layers in CNNs?**

- A. Increases parameter count
- B. Reduces computational complexity **(Answer)**
- C. Adds more non-linearity
- D. Performs normalization

---

#### **27. What does the term "feature map" refer to in CNNs?**
- A. A mapping of input labels
- B. The output of a convolutional layer **(Answer)**
- C. The input dimensions of an image
- D. A normalized image
---

#### **28. Which LLM focuses on efficiency by balancing model size and
training data?**
- A. GPT-3
- B. T5
- C. Chinchilla **(Answer)**
- D. PaLM

---

#### **29. In logistic regression, what happens when multicollinearity exists

between features?**
- A. Improves accuracy
- B. Causes instability in coefficients **(Answer)**
- C. Reduces training time
- D. Optimizes regularization

---

#### **30. Which algorithm uses value-based methods to learn from rewards in
reinforcement learning?**
- A. Gradient Boosting
- B. Q-Learning **(Answer)**
- C. k-Means Clustering
- D. Support Vector Machines

---

#### **31. Which LLM has 176 billion parameters and supports multilingual
applications?**
- A. Bloom **(Answer)**
- B. BERT
- C. GPT-2
- D. OpenAI Codex

---

#### **32. What does positional encoding help retain in transformer

architectures?**
- A. Spatial relations of images
- B. Sequence information in text **(Answer)**
- C. Activation patterns
- D. Loss convergence

---

#### **33. Which type of neural network is best suited for sequential data?**
- A. CNN
- B. RNN **(Answer)**
- C. Transformer
- D. Logistic Regression

---

#### **34. Which step in the ML project lifecycle involves defining success
metrics?**
- A. Data Collection
- B. Define Objective **(Answer)**
- C. Model Evaluation
- D. Feature Engineering

---

#### **35. Which operation helps in dimensionality reduction in CNNs?**

- A. Convolution
- B. Pooling **(Answer)**
- C. Backpropagation
- D. Dropout

---

#### **36. What does the softmax function output for multi-class classification?
**
- A. Binary labels
- B. Probabilities for each class **(Answer)**
- C. Decision boundaries
- D. Loss values

---

#### **37. Which model architecture does T5 use to solve NLP tasks?**
- A. Encoder-only
- B. Decoder-only
- C. Encoder-Decoder **(Answer)**
- D. Recurrent Neural Network

---

#### **38. What is the primary role of the output layer in CNNs?**
- A. Extract features
- B. Generate predictions **(Answer)**
- C. Reduce dimensionality
- D. Normalize inputs

---
#### **39. What metric is commonly used to measure the performance of a
binary classification model?**
- A. Mean Absolute Error
- B. R-squared
- C. Accuracy **(Answer)**
- D. Silhouette Score

---

#### **40. What process updates weights in neural networks during training?**
- A. Forward Propagation
- B. Backpropagation **(Answer)**
- C. Feature Engineering
- D. Pooling

---

#### **41. Which LLM introduced Mixture-of-Experts architecture for

scalability?**
- A. Bloom
- B. DeepSeek **(Answer)**
- C. LLaMA
- D. GPT-2

---

#### **42. Which neural network concept enables feature extraction by sliding
filters across input data?**
- A. Attention Mechanism
- B. Convolution **(Answer)**
- C. Pooling
- D. Embeddings
---

#### **43. Which dataset transformation enables logistic regression for binary
classification?**
- A. Label normalization
- B. Thresholding continuous target values **(Answer)**
- C. Image segmentation
- D. Dimensional reduction

---

#### **44. What happens when data drift occurs in ML models?**

- A. Model accuracy improves
- B. Data distribution changes over time **(Answer)**
- C. Training time decreases
- D. Loss function changes

---

#### **45. Which element is not part of logistic regression's training process?**
- A. Sigmoid function
- B. Loss function
- C. Convolutional layers **(Answer)**
- D. Optimization (Gradient Descent)

---

#### **46. Which optimization algorithm is widely used in transformers for

faster convergence?**
- A. Adam **(Answer)**
- B. SGD
- C. RMSprop
- D. Momentum
---

#### **47. Which architecture introduced self-attention for processing

sequences efficiently?**
- A. CNN
- B. Transformer **(Answer)**
- C. RNN
- D. DeepSeek

---

#### **48. In logistic regression, what does regularization help prevent?**

- A. Overfitting **(Answer)**
- B. Underfitting
- C. Parameter expansion
- D. Feature extraction

---

#### **49. Which neural network is best suited for tasks like image
classification?**
- A. RNN
- B. CNN **(Answer)**
- C. Transformer
- D. Logistic Regression

---

#### **50. What is the output of a convolution operation in CNNs?**

- A. Filter
- B. Feature Map **(Answer)**
- C. Decision boundary
- D. Loss value

---