Let's learn about some of the dimensionality reduction techniques that feature prominently in neuroscience literature.
This activity is not about getting every question right. It's about exploring the impact each technique would have on a project, and how different techniques can empower or mislead researchers—you are not expected to know the techniques featured, and this will not teach you how to implement them on a technical level.
Principal Component Analysis (PCA)
Often used for fMRI pattern analysis & variance-based feature extraction
Independent Component Analysis (ICA)
Common for EEG/MEG source separation and/or artifact removal
Linear Discriminant Analysis (LDA)
Commonly used to decode labeled states in neural data.
Uniform Manifold Approximation and Projection (UMAP)
Often used in studies featuring scRNA-seq and/or complex population activity
Non-negative Matrix Factorization (NMF)
Commonly used in calcium-imaging analysis and fMRI brain mapping
Over the next few screens, you'll be presented with various research scenarios that each feature a question in need of an answer. You are tasked with picking a dimensionality reduction technique that might answer that question.