Journal

Independent Component Analysis (ICA)

machine learning

math

PCA finds an orthogonal basis that decorrelates the data — after projection, all pairwise covariances are zero. But zero covariance does not imply statistical independence in…

Principal Component Analysis (PCA)

machine learning

math

For a set of data points in high-dimensional space, PCA finds a sequence of directions (eigenvectors of the covariance matrix) such that:

Thalamic Activation of V1 at the Single-Synapse Level

Neuroscience

paper

Chen, Kloos et al., Science 391, 1349 (2026)

DINO V1: Self-Supervised Vision Transformers

AI

paper

DINO can be viewed as an online clustering method, where each output dimension acts as a prototype. Instead of directly matching features, the model learns to assign…