Closed scottgigante-immunai closed 1 year ago
The high-dimensional PCA does well in global metrics, but poorly in local knn metrics. This baseline (200-dimensional laplacian eigenmaps) should perform well with local knn ranking metrics.
Tests passing at https://tower.nf/orgs/openproblems-bio/workspaces/openproblems-bio/watch/nnZQxJ2esb0PE
scottgigante-immunai
commented
1 year ago
The high-dimensional PCA does well in global metrics, but poorly in local knn metrics. This baseline (200-dimensional laplacian eigenmaps) should perform well with local knn ranking metrics.