Deep Learning Question Paper Sem 7 AI&DS Mumbai University

Deep Learning Question Paper Sem 7 AI&DS Mumbai University, Question’s are given with module and topic

Q1)

a) Design AND gate using Perceptron. (Module 1 [Multilayer Perceptrons (MLPs){1}])
b) Suppose we have N input-output pairs. Our goal is to find the parameters that predict the output y from the input x according to some function y = xw. Calculate the sum-of squared error function E between predictions y and inputs x. The parameter w can be determined iteratively using gradient descent. For the calculated error function E, derive the gradient descent update rule w ← w – α. (Module 1 [Gradient Descent{2}])
c) Explain dropout. How does it solve the problem of overfitting? (Module 2 [Regularization: Dropout{2}])
d) Explain denoising autoencoder model. (Module 3 [Autoencoders: Denoising Autoencoders{2}])
e) Describe sequence learning problem. (Module 5 [Recurrent Neural Networks: Sequence Learning Problem{1}])

Q2)

a) Explain Gated Recurrent Unit in detail. (Module 5 [Recurrent Neural Networks: Gated Recurrent Unit (GRU){1}])
b) What is an activation function? Describe any four activation functions. (Module 2 [Training: Activation Functions{2}])

Q3)

a) Explain CNN architecture in detail. Suppose we have an input volume of 32323 for a layer in CNN and there are ten 5*5 filters with stride 1 and pad 2; calculate the number of parameters in this layer of CNN. (Module 4 [Convolutional Neural Networks: CNN Architecture{2}])
b) Explain early stopping, batch normalization, and data augmentation. (Module 2 [Regularization: Early Stopping, Batch Normalization, Data Augmentation{2}])

Q4)

a) Explain RNN architecture in detail. (Module 5 [Recurrent Neural Networks: Architecture{1}])
b) Explain the working of Generative Adversarial Network. (Module 6 [Generative Adversarial Networks: Architecture{2}])

Q5)

a) Explain Stochastic Gradient Descent and momentum-based gradient descent optimization techniques. (Module 2 [Optimization: Stochastic Gradient Descent, Momentum-Based Gradient Descent{2}])
b) Explain LSTM architecture. (Module 5 [Recurrent Neural Networks: LSTM{2}])

Q6)

a) Describe LeNET architecture. (Module 4 [Modern Deep Learning Architectures: LeNET{1}])
b) Explain vanishing and exploding gradient in RNNs. (Module 5 [Recurrent Neural Networks: Vanishing and Exploding Gradients{1}])

Q1)

a) What are Feed Forward Neural Networks? (Module 1 [Feedforward Neural Networks{1}])
b) Explain Gradient Descent in Deep Learning. (Module 1 [Gradient Descent{1}])
c) Explain the dropout method and its advantages. (Module 2 [Regularization: Dropout{1}])
d) What are Undercomplete Autoencoders? (Module 3 [Undercomplete Autoencoder{1}])
e) Explain Pooling operation in CNN. (Module 4 [Convolutional Neural Networks: Pooling Layer{1}])

Q2)

a) What are the Three Classes of Deep Learning, explain each? (Module 1 [Deep Networks: Three Classes of Deep Learning{1}])
b) Explain the architecture of CNN with the help of a diagram. (Module 4 [Convolutional Neural Networks: CNN Architecture{1}])

Q3)

a) What are the different types of Gradient Descent methods, explain any three of them. (Module 2 [Optimization: Gradient Descent, Stochastic Gradient Descent, Mini Batch Gradient Descent{1}])
b) Explain the main components of an Autoencoder and its architecture. (Module 3 [Autoencoders: Unsupervised Learning{1}])

Q4)

a) Explain LSTM model, how it overcomes the limitation of RNN. (Module 5 [Recurrent Neural Networks: LSTM{1}])
b) What are the issues faced by Vanilla GAN models? (Module 6 [Generative Adversarial Networks: Architecture{1}])

Q5)

a) What are L1 and L2 regularization methods? (Module 2 [Regularization: L1, L2 Regularization{1}])
b) Explain any three types of Autoencoders. (Module 3 [Autoencoders: Denoising, Sparse, Contractive{1}])

Q6)

a) What is the significance of Activation Functions in Neural Networks, explain different types of Activation functions used in NN. (Module 2 [Training: Activation Functions{1}])
b) What are Generative Adversarial Networks, comment on their applications. (Module 6 [Generative Adversarial Networks: Applications{1}])

Table of Contents