Unsupervised Learning - Vector Quantisation

What type of data is suitable for unsupervised learning?

Unsupervised learning is suitable for unlabeled data, where the algorithm learns patterns and structures from the data without predefined categories or labels.

Name one algorithm commonly used for vector quantization.

K-means clustering is a common algorithm used for vector quantization.

What is another algorithm used for vector quantization besides K-means?

Self-Organizing Maps (SOM) are also used for vector quantization.

How do K-means and K-medoids differ?

K-means uses the mean of data points to form clusters while K-medoids uses actual data points (medoids) as center points.

What should you consider when using hierarchical clustering?

Hierarchical Clustering should be used when the data has a nested structure and you want to understand the relationships between clusters at various levels.

What is the difference between density-based spatial clustering of applications with noise (DBSCAN) and K-means?

DBSCAN can identify clusters of varying shapes, sizes and can handle noise, while K-means requires spherical clusters of similar sizes.

In which scenarios would Principal Component Analysis (PCA) be used?

PCA is used for dimensionality reduction, especially when dealing with high-dimensional data to improve computational efficiency and visualization.

What is the purpose of Gaussian Mixture Models (GMM) in unsupervised learning?

GMM’s are used to model data as a mixture of multiple Gaussian distributions and can capture the underlying distribution of the data.

How can R be used in relation to unsupervised learning?

R can be used to create scripts and functions for implementing unsupervised learning algorithms like K-means, PCA, or clustering methods.

What does evaluating unsupervised learning algorithms involve?

Evaluating unsupervised learning algorithms involves assessing cluster quality, silhouette scores and the ability to generalize new data.