Natural Language Processing
CSE · Final · Fall 24
Paper 1 of 2Batch 61Switch
Batch 61
Paper 1 of 2
Read the questions as text
Read off the PDF by AI, so it may have mistakes. The PDF is the original.
Daffodil International University
Faculty of Science & Information Technology
Department of Computer Science and Engineering
Final Examination, Fall-2024
Course Code: CSE445, Course Title: Natural Language Processing
Level: 4 Term: 1 Batch: 61
Time: 2 Hours
Marks: 40
Answer ALL Questions
[The figures in the right margin indicate the full marks and corresponding course outcomes. All portions of each question must be answered sequentially.]
1. a) Describe topic modeling method with an example? Is topics modeling and topic distribution are same, Justify your answer? [5] CO2
1. b) Explain the components of Recurrent Neural Network (RNN) architecture with a diagram. How does an RNN work? [5]
2. a) How does speech analysis provide insights into human behavior and communication, and what factors affect its accuracy and effectiveness? [5] CO2
2. b) What tools are available for feature extraction from speech data, and which methods are used for noise reduction? [5]
3. Consider the following dataset:
Text | Vector Representation | Label
"The movie was fantastic!" | [1, 0, 0, 1, 0, 1, 0, 0, 1] | 1
"I didn't enjoy the film." | [0, 1, 1, 0, 1, 0, 1, 1, 0] | 0
"Amazing plot and acting!" | [1, 1, 0, 1, 0, 1, 0, 0, 0, 1] | 1
"A terrible waste of time." | [0, 0, 1, 0, 1, 0, 1, 1, 1, 0] | 0
Input Layer -> Hidden Layer Weights (W1):
W1 = [0.3, -0.2, 0.6, 0.1, -0.4, 0.5, 0.2, -0.1, 0.7, -0.3]
[Figure: A weight diagram]
A scalar bias for the hidden layer neuron b1 = 0.2
Hidden Layer -> Output Layer Weight (W2):
W2 = [0.8]
A scalar bias for the output layer neuron b2 = -0.1
3. a) Draw the neural network based on the input vectors and weight matrices. [2] CO3
3. b) Predict the class for the vector [0,1,1,0,1,0,0,1,1,0]. Assume ReLU as the activation function for the hidden layer and SoftMax for the classification function. [6]
3. c) Calculate the error using mean square error. [2]
4. Consider the following Neural Network:
[Figure: A neural network diagram with input nodes X1-0.35, X2-0.45, weights w1=0.1, w2=0.2, w3=0.3, w4=0.4, hidden nodes h11, h12 with bias b1=1, weights w5=0.5, w6=0.6, w7=0.7, w8=0.8, hidden nodes h21, h22 with bias b2=1, weights w9=0.9, w10=0.99, output node out=0.75 with bias b3=1]
When Activation and classification Function = Swish(x) = x / (1 + e^-x) [10] CO3
If the learning rate is n = 0.01 calculate adjusted weight for w9, w5, w10.