This project investigates how ANN methods deform the geometric structure of image embedding neighborhoods when compared to an exact k-NN search.
nearest-neighbor-search approximate-nearest-neighbor-search manifold-learning similarity-search hnsw high-dimensionality image-embeddings local-intrinsic-dimensionality intrinsic-dimensionality neighbor-hood-geometry barycenter-shift
-
Updated
Dec 22, 2025 - Jupyter Notebook