Scribd
Upload
9 views2 pages

Mid-Sem Exam Questions on ML & CNNs

The document consists of sample questions for a mid-semester exam covering various topics in machine learning, including Bayesian learning, SVM algorithms, CNN implementation in TensorFlow, and reinforcement learning methods. It addresses concepts such as overfitting, feature extraction, data preprocessing, and model evaluation. Additionally, it explores advanced architectures like Inception networks and their advantages over traditional models.

Uploaded by

vishalkushwaha9752
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
9 views2 pages

Mid-Sem Exam Questions on ML & CNNs

The document consists of sample questions for a mid-semester exam covering various topics in machine learning, including Bayesian learning, SVM algorithms, CNN implementation in TensorFlow, and reinforcement learning methods. It addresses concepts such as overfitting, feature extraction, data preprocessing, and model evaluation. Additionally, it explores advanced architectures like Inception networks and their advantages over traditional models.

Uploaded by

vishalkushwaha9752
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd

Sample Question for mid sem

1) Define Bayesian learning and discuss its role in model uncertainty and decision-making
in machine learning.
2) In what ways does Bayesian learning provide a different perspective on overfitting
compared to traditional ML approaches?
3) How does the SVM algorithm classify data, and what is the role of hyperplanes in this
process?
4) Compare and contrast SVMs and Bayesian models in handling high-dimensional feature
spaces.
5) Given a text classification task, explain why SVM might be more efficient than a neural
network.
6)  List the steps to implement a basic CNN model in TensorFlow, including data
preprocessing and model evaluation.
7)  Explain the importance of data preprocessing and augmentation when training CNNs
on image datasets.
8)  Describe how TensorFlow supports CNN training, including layers and optimization
tools.
9)  Provide a code snippet for defining a simple CNN model in TensorFlow or Keras.
10) Differentiate between convolutional and pooling layers in a CNN. How do they impact
feature extraction?
11) What is the function of a loss layer in neural networks, and how does it guide model
learning?
12) Explain the Actor-Critic model in reinforcement learning and how it balances policy and
value learning.
13) Compare Q-learning and Actor-Critic methods in reinforcement learning. In which
scenarios might each be preferred?
14) What are Long Short-Term Memory (LSTM) networks, and why are they effective for
sequence data like text and speech?
15) Define the BLEU score and describe how it is used to evaluate language generation tasks
such as machine translation.
16) Compare the structure and feature extraction capabilities of Convolutional Neural
Networks (CNNs) versus traditional Artificial Neural Networks (ANNs).
17) What key innovations does the Inception network introduce over traditional CNN
architectures, and how do they affect performance and complexity?
18) Why are CNNs preferred over ANNs for image data? Justify with two technical reasons.
19) Design a scenario where Inception networks outperform basic deep learning models in
terms of efficiency and accuracy.
20) Explain how model selection affects the training performance and generalization ability
of a CNN in TensorFlow.

You might also like