Corey J. Nolet

[FEA] implement chunking in hdbscan to support smaller GPUs

1

Once we are able to run a cross-component 1nn to reduce the memory footprint of connecting the disconnected components in the MST, we should be able to provide the option...

[FEA] UMAP to support more `target_metric` options than `categorical` and `eucliean`

1

This is a feature request to support different metrics for `target_metric` in UMAP other than just the initial settings of `categorical` and `euclidean`. Ref #4816.

[FEA] Investigate approximate nearest neighbors in UMAP/TSNE/HDBSCAN

1

The computation of the all-nearest-neighbors KNN is the bottleneck in UMAP and TSNE and both of these algorithms can get away with using approximate neighbors as the number of points...

Getting `concurrent.futures._base.CancelledError` from simple binary tree built from futures

24

I'm getting the following exception from a binary tree that I'm building from futures (based largely on an example in the dask docs that uses delayed instead of futures). Here's...

JObject appears to use a List(JField) underneath so constant time \() is not possible

2

I'm on a project which uses json4s with a heavy amount of get operations for retrieving values for keys. This project is processing data on the order of 10's of...

The Fluo table has a rather complex structure of prefix masking in the timestamp of the key as well as byte arrays in the value that mean different things based...