srowen commented on a change in pull request #26415: [SPARK-18409][ML] LSH
approxNearestNeighbors should use approxQuantile instead of sort
URL: https://github.com/apache/spark/pull/26415#discussion_r348886645
##########
File path: mllib/src/main/scala/org/apache/spark/ml/feature/LSH.scala
##########
@@ -137,14 +139,21 @@ private[ml] abstract class LSHModel[T <: LSHModel[T]]
val hashDistUDF = udf((x: Seq[Vector]) => hashDistance(x, keyHash),
DataTypes.DoubleType)
val hashDistCol = hashDistUDF(col($(outputCol)))
- // Compute threshold to get exact k elements.
- // TODO: SPARK-18409: Use approxQuantile to get the threshold
- val modelDatasetSortedByHash =
modelDataset.sort(hashDistCol).limit(numNearestNeighbors)
- val thresholdDataset = modelDatasetSortedByHash.select(max(hashDistCol))
- val hashThreshold = thresholdDataset.take(1).head.getDouble(0)
-
- // Filter the dataset where the hash value is less than the threshold.
- modelDataset.filter(hashDistCol <= hashThreshold)
+ // Compute threshold to get around k elements.
Review comment:
From ALS? OK I could see that being an efficient alternative, to get top k
per partition and merge them to a final top k -- if k isn't big. The existing
impl works on keyed data but a simplified version would work here. I think that
could be a valid further improvement.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
