Machine Learning - Similarity Based Learning Quiz

1. What type of learning algorithm is K-Nearest Neighbors (K-NN)?

• Unsupervised Learning
  

• Supervised Learning
  

• Reinforcement Learning
  

• Deep Learning

Why is K-NN referred to as a lazy learning algorithm?

• Because it does not use training data

• Because it builds the model during the training phase

• Because it stores training data and delays processing until prediction

• Because it learns from mistakes

Which distance metric is most commonly used in K-NN?

• Hamming Distance

• Euclidean Distance

• Cosine Similarity

• Manhattan Distance

Which of the following is a drawback of similarity-based learning?

• Requires large memory to store data

• Requires pre-training of data

• Cannot classify new data

• Builds abstract models

In weighted K-NN, how are the weights assigned to neighbors?

• Randomly

• Based on class frequency

• Inversely proportional to distance

• Based on their order

Which of the following is an instance-based learning algorithm?

• Decision Trees

• Support Vector Machines

• K-Nearest Neighbors

• Naive Bayes

What is the main idea behind Weighted K-Nearest Neighbors (K-NN)?

• All neighbors have equal voting power

• The nearest neighbor alone decides the class

• Neighbors contribute based on their distance from the query point

• Only labeled data is used

What does a Centroid Classifier use to represent each class?

• The most frequent label

• The median of the data points

• A decision tree

• The average (mean) of its training points

Which of the following statements is true about Centroid Classifier?

• It needs to store the entire training set for prediction

• It uses the K-nearest neighbors for classification

• 
  It summarizes each class with a single mean vector

• It only works with binary classification

How can the performance of K-NN be improved?

• By increasing the number of irrelevant features

• By reducing the training data

• By normalizing the input features

• By decreasing the value of k to 1