Instance-Based Learning QUIZ (MCQ QUESTIONS AND ANSWERS)

Faster training time

Faster prediction time

Better handling of noisy data

Lower memory requirements

By using a more complex similarity measure

By using a simpler similarity measure

By using indexing structures to efficiently search for nearest neighbors

By using more nearest neighbors for classification

k-Nearest Neighbors is an instance-based learning algorithm, while k-Means is a clustering algorithm.

k-Nearest Neighbors is a clustering algorithm, while k-Means is an instance-based learning algorithm.

k-Nearest Neighbors uses a distance metric, while k-Means uses a similarity measure.

k-Nearest Neighbors uses a similarity measure, while k-Means uses a distance metric.

Linear Discriminant Analysis (LDA)

Logistic Regression

k-Nearest Neighbors

Perceptron

It reduces the performance of the algorithm

It has no effect on the performance of the algorithm

It improves the performance of the algorithm

It depends on the type of similarity measure used

By setting a threshold on the distance to the k nearest neighbors

By setting a threshold on the number of nearest neighbors

By setting a threshold on the similarity measure

By setting a threshold on the classification accuracy

A memory of specific instances in the training data

A memory of the general patterns in the training data

A memory of the similarity measures used during training

A memory of the classification rules learned during training

Curse of dimensionality

Handling missing values

Sensitivity to noisy instances

Sensitivity to irrelevant features

Decision Trees

k-Nearest Neighbors

Support Vector Machines

Neural Networks

It can handle missing values

It can handle imbalanced datasets

It reduces the effect of noisy instances

It increases the computation time during prediction

The algorithm is biased towards the majority class

The algorithm is biased towards the minority class

The algorithm is not sensitive to class imbalance

The algorithm cannot handle imbalanced datasets

To reduce the number of instances in the training data

To reduce the number of features in the training data

To reduce the complexity of the instance-based learning algorithm

To reduce the computation time during prediction

Cross-validation

Grid search

Random search

All of the above

k-Nearest Neighbors

Learning Vector Quantization (LVQ)

Self-Organizing Maps (SOM)

Radial Basis Function Networks (RBFN)

When the data is linearly separable

When the data has complex non-linear patterns

When the data has a large number of features

When the data has a large number of instances

By removing instances with missing values

By imputing missing values using the mean or median

By using similarity measures that can handle missing values

All of the above

It improves the performance of the algorithm

It has no effect on the performance of the algorithm

It worsens the performance of the algorithm

It depends on the type of similarity measure used

Instance-based learning uses training data to create a model, while model-based learning relies on instances for classification.

Instance-based learning does not create a model, while model-based learning creates a model from the training data.

Instance-based learning only works on numeric data, while model-based learning works on both numeric and categorical data.

Instance-based learning is a type of supervised learning, while model-based learning is a type of unsupervised learning.

Euclidean distance

Cosine similarity

Pearson correlation coefficient

Linear regression

To give more importance to closer instances when determining the class of a new data point

To give more importance to further instances when determining the class of a new data point

To give equal importance to all instances when determining the class of a new data point

To give more importance to instances with more features when determining the class of a new data point

By selecting the majority class for each label among the k nearest instances

By selecting the majority class among the k nearest instances for each label

By selecting the class with the highest probability for each label among the k nearest instances

By selecting the class with the highest probability among the k nearest instances for each label

Overfitting

Underfitting

Increased computation time

Decreased accuracy

By selecting the majority class among the k nearest instances

By calculating the mean of the target values of the k nearest instances

By calculating the mode of the target values of the k nearest instances

By calculating the median of the target values of the k nearest instances

Feature selection

Feature extraction

Dimensionality reduction

All of the above

Calculate the distance between the new data point and all training instances

Select the k nearest instances

Train a model using the k nearest instances

Determine the majority class among the k nearest instances

Overfitting

Underfitting

Increased computation time

Decreased accuracy

They are computationally expensive during training

They are computationally expensive during prediction

They require less memory than model-based algorithms

They are not affected by the curse of dimensionality

The number of instances in the dataset

The number of features in the dataset

The number of nearest instances considered for classification

The number of possible labels for classification

Manhattan distance

Euclidean distance

Minkowski distance

Both A and B

Decision Trees

k-Nearest Neighbors (k-NN)

Support Vector Machines

Linear Regression