commit 3a73b77e4187d51ded0d0a5b81d32d3a1f407156
author Frederic Zoepffel <f.zoepffel@gmail.com> Sat Sep 14 15:29:08 2024 +0200
committer Matthias Boehm <mboehm7@gmail.com> Sat Sep 14 15:42:06 2024 +0200
tree cee3b1bbc98627ffa4555d863e800d84509f1af1
parent 726d21d08aa417764123221e2f5ae95ff92bb4f9

[SYSTEMDS-3696] Minor robustness fix and pruning flags

Closes #2107.

scripts/builtin/incSliceLine.dml

diff --git a/scripts/builtin/incSliceLine.dml b/scripts/builtin/incSliceLine.dml
index 212d12e..ff74196 100644
--- a/scripts/builtin/incSliceLine.dml
+++ b/scripts/builtin/incSliceLine.dml
@@ -52,6 +52,8 @@
 # prevTK           previous top-k slices (for incremental updates)
 # prevTKC          previous top-k scores (for incremental updates)
 # encodeLat        flag for encoding output lattice for less memory consumption
+# pruningStrat     flag for disabling certain pruning strategies
+#                  (0 all, 1 all exact (score and size), 2 no score, 3 no size, 4 none)
 # ---------------------------------------------------------------------------------------
 #
 # OUTPUT:
@@ -99,8 +101,9 @@
       + " -- see documentation for more details.");
   }
 
-  disableIncScorePruning = (pruningStrat == 1 | pruningStrat == 3);
-  disableIncSizePruning = (pruningStrat >= 2);
+  disableIncScorePruning = (pruningStrat == 2 | pruningStrat == 4);
+  disableIncSizePruning = (pruningStrat >= 3);
+  disableIncApproxPruning = (pruningStrat >= 1)
 
   t1 = time();
 
@@ -183,9 +186,9 @@
   # create and score basic slices (conjunctions of 1 feature)
   maxsc = getMaxScoreAllFeatures(nrow(X2), ncol(X2), prevLattice, metaPrevLattice,
      prevStats, encodeLat, differentOffsets, alpha, eAvg, prevFoffb, prevFoffe, foffb, foffe);
-  maxscub = getMaxChangedScoreAllFeatures(nrow(X2), ncol(X2),
-     addedX2, removedX2, addedE, removedE, prevLattice, metaPrevLattice, prevStats,
-     encodeLat, differentOffsets, alpha, eAvg, minSup, prevFoffb, prevFoffe, foffb, foffe);
+  maxscub = getMaxChangedScoreAllFeatures(nrow(X2), ncol(X2), addedX2, removedX2,
+     addedE, removedE, prevLattice, metaPrevLattice, prevStats, encodeLat, differentOffsets,
+     alpha, eAvg, minSup, prevFoffb, prevFoffe, foffb, foffe, disableIncApproxPruning);
   [S, R, selCols] = createAndScoreBasicSlicesInc(X2, changedX2, prevTK2, totalE, changedE,
      eAvg, eAvgOld, eAvgNew, minSup, alpha, minsc, maxsc, maxscub, verbose, disableIncScorePruning);
 
@@ -328,7 +331,10 @@
 
   # b) unchanged pruning
   # (valid to prune feature if its previous max score was negative or below minsc)
-  selCols2 = selCols & (ncCnts > 0 | maxsc > max(0, minsc));
+  selCols2 = selCols;
+  if( !disableIncScorePruning ) {
+    selCols2 = selCols & (ncCnts > 0 | maxsc > max(0, minsc));
+  } 
 
   if( verbose ) {
     n = as.integer(sum(selCols));
@@ -370,7 +376,11 @@
       print("incSliceLine: dropping "+drop+"/"+n+" features below minSore = "+minsc+".");
     }
     cix = removeEmpty(target=attr, margin="rows", select=selCols3);
-    selCols = table(cix, 1, n2, 1);
+    if(sum(cix) != 0) {
+      selCols = table(cix, 1, n2, 1);
+    }else {
+      selCols = matrix(0, n2, 1);
+    }
   }
   else {
     selCols = selCols2;
@@ -743,11 +753,12 @@
   Matrix[Double] removedX2, Matrix[Double] addedE, Matrix[Double] removedE,
   List[Unknown] prevLattice, List[Unknown] metaPrevLattice, List[Unknown] prevStats,
   Boolean encodeLat, Boolean differentOffsets, Double alpha, Double eAvg, Double minSup,
-  Matrix[Double] prevFoffb, Matrix[Double] prevFoffe, Matrix[Double] foffb, Matrix[Double] foffe)
+  Matrix[Double] prevFoffb, Matrix[Double] prevFoffe, Matrix[Double] foffb, Matrix[Double] foffe,
+  Boolean disableIncApproxPruning)
   return(Matrix[Double] maxscub)
 {
   maxscub = matrix(-Inf, numFeatures, 1);
-  if( length(prevLattice) > 0 & nrow(addedX2) < 0.05*numRows ) {
+  if( length(prevLattice) > 0 & nrow(addedX2) < 0.05*numRows & !disableIncApproxPruning ) {
     # compute upper bounds per feature for added subset
     ss = t(colSums(addedX2));
     se = t(t(addedE) %*% addedX2);