INTRODUCTION TO DEEP LEARNING
Module-wise 2 & 5 Mark Questions and Answers
MODULE – I : ML FUNDAMENTALS
Q1. What is Machine Learning? (2 Marks)
Machine Learning is a branch of Artificial Intelligence that enables a system to learn from data and
improve its performance without explicit programming.
Q2. What is Supervised Learning? (2 Marks)
Supervised learning is a type of machine learning where the model is trained using labeled data.
Q3. What is Gradient Descent? (2 Marks)
Gradient Descent is an optimization algorithm used to minimize the error by updating parameters step
by step.
Q4. What is Overfitting? (2 Marks)
Overfitting occurs when a model performs well on training data but poorly on unseen test data.
Q5. What is Mean Squared Error (MSE)? (2 Marks)
MSE is the average of squared differences between actual values and predicted values.
Q6. Explain Gradient Descent Algorithm. (5 Marks)
Gradient Descent is an optimization algorithm used to minimize the cost function.
Steps:
1. Initialize parameters randomly
2. Calculate predicted output
3. Compute error
4. Calculate gradient
5. Update parameters using learning rate
6. Repeat until minimum error is reached
Formula:
θ = θ − α ∂J(θ)/∂θ
Advantages:
• Simple and efficient
• Used in regression and neural networks
Q7. Explain Linear Regression. (5 Marks)
Linear Regression is a supervised learning algorithm used to predict continuous values.
Equation:
y = mx + c
Working:
• Finds best fit line
• Uses least squares method
• Error minimized using gradient descent
Applications:
• House price prediction
• Sales forecasting
Q8. Explain Confusion Matrix and Evaluation Metrics. (5 Marks)
A Confusion Matrix evaluates classification performance.
Terms:
• TP – True Positive
• TN – True Negative
• FP – False Positive
• FN – False Negative
Metrics:
• Precision = TP / (TP + FP)
• Recall = TP / (TP + FN)
• F1-Score = 2PR / (P + R)
MODULE – II : ADVANCED ML TECHNIQUES
Q1. What is SVM? (2 Marks)
Support Vector Machine is a supervised learning algorithm that finds the optimal hyperplane for
classification.
Q2. What is Kernel Trick? (2 Marks)
Kernel Trick converts non-linear data into higher dimensional space to make it linearly separable.
Q3. What is k-Means Clustering? (2 Marks)
k-Means is an unsupervised algorithm that groups data into k clusters based on similarity.
Q4. What is PCA? (2 Marks)
PCA is a dimensionality reduction technique used to reduce features while preserving important
information.
Q5. Explain Support Vector Machine with Kernel Trick. (5 Marks)
SVM finds a hyperplane with maximum margin between classes.
Kernel Trick:
Used for non-linear data.
Types of Kernels:
• Linear
• Polynomial
• RBF
Advantages:
• Works well with high-dimensional data
• Effective for small datasets
Q6. Explain k-Means Clustering Algorithm. (5 Marks)
Steps:
1. Choose k
2. Initialize centroids
3. Assign points to nearest centroid
4. Update centroids
5. Repeat until convergence
Applications:
• Customer segmentation
• Image compression
Q7. Explain Principal Component Analysis (PCA). (5 Marks)
Steps:
1. Standardize data
2. Compute covariance matrix
3. Find eigenvalues and eigenvectors
4. Select principal components
Advantages:
• Reduces complexity
• Improves performance
MODULE – III : DEEP LEARNING FUNDAMENTALS
Q1. What is Deep Learning? (2 Marks)
Deep Learning is a subset of machine learning that uses multi-layer neural networks.
Q2. What is Backpropagation? (2 Marks)
Backpropagation is an algorithm used to update weights in neural networks by propagating error
backward.
Q3. What is CNN? (2 Marks)
CNN is a deep learning model mainly used for image processing.
Q4. What is Vanishing Gradient Problem? (2 Marks)
It occurs when gradients become very small, slowing or stopping learning in deep networks.
Q5. Explain Backpropagation Algorithm. (5 Marks)
Steps:
1. Forward pass
2. Calculate error
3. Backward pass
4. Update weights
Advantages:
• Reduces error
• Improves accuracy
Q6. Explain CNN Architecture. (5 Marks)
Layers:
• Convolution layer
• ReLU layer
• Pooling layer
• Fully connected layer
Applications:
• Face recognition
• Image classification
Q7. Explain LSTM. (5 Marks)
Components:
• Forget gate
• Input gate
• Output gate
Advantages:
• Handles long-term dependencies
• Solves vanishing gradient problem
MODULE – IV : ADVANCED DL CONCEPTS
Q1. What is GAN? (2 Marks)
GAN is a deep learning model consisting of Generator and Discriminator used to generate new data.
Q2. What is Transformer? (2 Marks)
Transformer is a deep learning model based on attention mechanism mainly used in NLP.
Q3. What is Q-Learning? (2 Marks)
Q-Learning is a reinforcement learning algorithm that learns optimal actions using rewards.
Q4. Explain GAN Architecture. (5 Marks)
Components:
1. Generator
2. Discriminator
Working:
• Generator creates fake data
• Discriminator detects real or fake
• Both improve together
Applications:
• Image generation
• Data augmentation
Q5. Explain Transformer Architecture. (5 Marks)
Components:
• Self-Attention
• Multi-Head Attention
• Encoder–Decoder
Advantages:
• Parallel processing
• Handles long sequences