Machine Learning and AI Quiz

A retail company wants to predict whether a customer will make a purchase based on their browsing history, demographics, and past purchases. Which model will be most appropriate for this?

• Unsupervised clustering model

• Supervised regression model

• Supervised classification model

• Reinforcement learning model

• Large Language Model

What is the primary purpose of a loss function in machine learning?

• To prevent overfitting

• To measure the accuracy of the model on unseen data

• To quantify the difference between predicted and actual values

• To optimize the learning rate

Which of the following scenarios would benefit most from using a Convolutional Neural Network (CNN)?

• Predicting stock prices over time

• Classifying customer reviews as positive or negative

• Detecting tumors in medical X-ray images

• Generating human-like text responses

A social media platform wants to automatically group similar posts together without predefined categories. Which approach should they use?

• Supervised classification with labels

• Linear regression

• Unsupervised clustering

• Large Language Model

What is a key advantage of Deep Learning over traditional Machine Learning for complex tasks like image recognition?

• Requires less data for training

• Eliminates the need for feature engineering

• Easier to interpret model decisions

• Faster training times on large datasets

What distinguishes deep learning from traditional machine learning?

• Deep learning only works with unstructured data and machine learning works only with structured data

• Deep learning uses multiple layers to learn features and hierarchical representations

• Deep learning requires less data

• Deep learning is always more accurate

A video streaming platform wants to recommend movies based on viewing history. They notice that users who watch sci-fi movies also tend to watch documentaries about space. This pattern discovery is an example of:

• Supervised learning finding labeled patterns

• Unsupervised learning finding hidden associations

• Reinforcement learning through rewards

• Transfer learning from another domain

What is the benefit of using embeddings as numeric representation of words?

• Word embeddings are always high dimensional.

• Word embeddings capture semantic meaning of the words.

• Other approaches are computationally intensive

• Word embeddings require less training data.

Why might you increase the dimensionality of word embeddings from 100 to 300?

• To guarantee faster inference on CPU

• To allow the vectors to capture richer semantic and syntactic features

• To reduce the risk of overfitting on small corpora

• Because cosine similarity only works in ≥256 dimensions

Which type of attention allows each token to attend to all other tokens in the sequence simultaneously?

• Causal attention

• Cross attention

• Self-attention

• Sparse attention