DL basic

Question 1

Which NLP application would likely benefit MOST from an extensive stemming or lemmatization process?

Accepted Answer

Information retrieval (search)

Question 2

Which of the following is NOT a common type of layer in deep neural networks?

Accepted Answer

Backpropagation

Question 3

What is the main purpose of stemming in text preprocessing?

Accepted Answer

To convert words to their base or root form by removing affixes

Question 4

In TF-IDF, what does the IDF component accomplish?

Accepted Answer

It decreases the weight of words that appear in many documents

Question 5

Which of these neural network architectures is specifically designed for sequence data?

Accepted Answer

Recurrent Neural Networks (RNNs)

Question 6

What is the main advantage of TF-IDF over simple term frequency?

Accepted Answer

It downweights common words that appear in many documents

Question 7

Which of the following is a common test to evaluate the quality of word embeddings?

Accepted Answer

Word analogy tasks (e.g., "king" - "man" + "woman" ≈ "queen")

Question 8

What does a learning rate that is too high typically cause?

Accepted Answer

Unstable training or divergence

Question 9

In the Skip-gram model of Word2Vec, what is being predicted?

Accepted Answer

The context words from the center word

Question 10

What is a potential downside of removing stopwords for sentiment analysis?

Accepted Answer

It could remove important negations (e.g., "not good" becomes "good")

Question 11

Transfer learning in NLP refers to:

Accepted Answer

Using knowledge from one task to improve performance on another task

Question 12

In a neural network context, what does "backpropagation" refer to?

Accepted Answer

The algorithm for computing gradients and updating weights

Question 13

In a confusion matrix for binary classification, true positives are:

Accepted Answer

Examples correctly classified as the positive class

Question 14

In which domain has deep learning shown particularly remarkable success?

Accepted Answer

Computer vision

Question 15

Why is stopword removal often performed in NLP tasks?

Accepted Answer

remove words that occur too frequently to be informative e.g., "the", "is"

Question 16

What is a limitation of the Bag-of-Words model?

Accepted Answer

It loses word order information

Question 17

What is the purpose of dropout in neural networks?

Accepted Answer

regularize the network prevent overfitting by randomly deactivating neurons

Question 18

How would you represent the sentence "The cat sat on the mat" using a Bag-of-Words approach?

Accepted Answer

{"the": 2, "cat": 1, "sat": 1, "on": 1, "mat": 1}

Question 19

Which of these is NOT a common application of deep learning in NLP?

Accepted Answer

Database indexing

Question 20

Which of the following best describes Word2Vec?

Accepted Answer

technique that represents words as dense vectors in continuous vector space

Question 21

What is the dimensionality of a typical word embedding vector?

Accepted Answer

Between 50 and 300 dimensions

Question 22

Which activation function is commonly used in the output layer for binary classification?

Accepted Answer

Sigmoid

Question 23

Which technology forms the foundation of modern voice assistants like Siri and Alexa?

Accepted Answer

DL models for speech recognition and natural language understanding

Question 24

Which of these would NOT be considered a challenge in tokenization?

Accepted Answer

Converting text to lowercase

Question 25

Which metric would be most important to maximize when classifying rare diseases where false negatives are very costly?

Accepted Answer

Recall

Question 26

How does lemmatization differ from stemming?

Accepted Answer

it considers the context and part of speech to convert to proper base forms

Question 27

What type of problem is best solved using the Mean Squared Error (MSE) loss function?

Accepted Answer

Regression

Question 28

Which of these is NOT a common text preprocessing step in NLP?

Accepted Answer

Vector quantization

Question 29

Which loss function is most appropriate for multi-class classification?

Accepted Answer

Categorical cross-entropy

Question 30

Tokenization refers to:

Accepted Answer

Breaking text into meaningful units like words or subwords

Question 31

If a model has high precision but low recall for the positive class, this indicates:

Accepted Answer

model rarely predicts the positive class but usually correct when it does

Question 32

If a term has a high TF-IDF score in a document, this indicates:

Accepted Answer

term is relatively common in this document but rare across all documents

Question 33

Which statement best describes the relationship between deep learning and traditional machine learning?

Accepted Answer

subset of machine learning that uses neural networks with many layers

Question 34

What does the ReLU activation function return for negative inputs?

Accepted Answer

0

DL basic

Questions 34

DL basic

Questions 34