week 7 -chatgpt Flashcards
(11 cards)
What is the difference between feature selection and feature extraction?
Feature selection chooses a subset of existing features, while feature extraction creates new features from original data to improve classification.
What is the goal of Principal Component Analysis (PCA)?
To reduce dimensionality by projecting data onto directions (principal components) that maximize variance.
Is PCA a supervised or unsupervised method?
Unsupervised — it does not use class labels, only the variance structure of the data.
What does Linear Discriminant Analysis (LDA) optimize?
It maximizes the ratio of between-class variance to within-class variance to achieve better class separation.
How is LDA different from PCA?
LDA is supervised and focuses on class separation, while PCA is unsupervised and focuses on variance.
What does Independent Component Analysis (ICA) aim to find?
A representation of data where components are statistically independent, not just uncorrelated like PCA.
When does ICA give similar results to PCA?
When the data is Gaussian, since uncorrelated implies independence for Gaussian distributions.
What is the idea behind random projections for feature extraction?
To map data into a higher-dimensional space using random weights and a nonlinearity, where it may become linearly separable.
What is the main idea of sparse coding?
To represent input data using a small number of active features from a larger dictionary, encouraging interpretability and efficiency.
How is sparse coding different from PCA or ICA?
Sparse coding encourages sparsity in the feature vector, while PCA/ICA focus on orthogonal or independent directions.
Why might you use feature extraction instead of manual feature engineering?
To automatically discover transformations that improve model performance without requiring domain-specific knowledge.
