UMAP is faster than t-SNE

t-SNE preserves local structure

t-SNE preserves local structure by converting distances to probabilities and minimizing Kullback-Leibler divergence

Proximal gradient methods for learning

Proximal gradient descent efficiently handles non-differentiable L1 regularization by combining gradient descent with a proximity operator

the curse of dimensionality makes nearest neighbor search unreliable

High dimensionality dilutes data density, making nearest neighbors less distinct and search unreliable

batch size affects generalization: larger batches find sharper minima

Larger batch sizes lead to sharper minima, enhancing generalization by providing more accurate gradient estimates

Locality-sensitive hashing

Locality-sensitive hashing (LSH) hashes similar items into the same buckets

LoRA (machine learning)

LoRA uses r << d for efficient adaptation