naive frequency pruning

DillonZChen · Jan 23, 2025 · e2e872d · e2e872d
1 parent 5c68993
commit e2e872d
Show file tree

Hide file tree

Showing 4 changed files with 53 additions and 14 deletions.
diff --git a/include/feature_generation/features.hpp b/include/feature_generation/features.hpp
@@ -141,11 +141,11 @@ namespace feature_generation {
 
     std::set<int> prune_collapse_layer(int iteration, std::vector<std::vector<int>> &cur_colours);
     std::set<int> prune_collapse_layer_x(int iteration,
-                                         const std::vector<graph::Graph> &graphs,
-                                         std::vector<std::vector<int>> &cur_colours);
+                                         const std::vector<graph::Graph> &graphs);
     std::set<int> prune_collapse_layer_y(int iteration,
-                                         const std::vector<graph::Graph> &graphs,
-                                         std::vector<std::vector<int>> &cur_colours);
+                                         const std::vector<graph::Graph> &graphs);
+    std::set<int> prune_collapse_layer_f(int iteration,
+                                         const std::vector<graph::Graph> &graphs);
     std::set<int> prune_maxsat(std::vector<Embedding> X);
     std::set<int> prune_maxsat_x(std::vector<Embedding> X, const int maxsat_iterations);
 

diff --git a/include/feature_generation/pruning_options.hpp b/include/feature_generation/pruning_options.hpp
@@ -11,8 +11,10 @@ namespace feature_generation {
     static const std::string COLLAPSE_ALL;
     static const std::string COLLAPSE_ALL_X;
     static const std::string COLLAPSE_LAYER;
-    static const std::string COLLAPSE_LAYER_X;
-    static const std::string COLLAPSE_LAYER_Y;
+    static const std::string COLLAPSE_LAYER_X;  // bfg2019
+    static const std::string COLLAPSE_LAYER_Y;  // maxsat
+    static const std::string COLLAPSE_LAYER_F;  // frequency count < 1% n_data
+    static const std::string COLLAPSE_LAYER_YF;  // Y then F
 
     static const std::vector<std::string> get_all();
   };

diff --git a/src/feature_generation/pruning/layer_pruners.cpp b/src/feature_generation/pruning/layer_pruners.cpp
@@ -10,9 +10,15 @@ namespace feature_generation {
     if (pruning == PruningOptions::COLLAPSE_LAYER) {
       to_prune = prune_collapse_layer(iteration, cur_colours);
     } else if (pruning == PruningOptions::COLLAPSE_LAYER_X) {
-      to_prune = prune_collapse_layer_x(iteration, graphs, cur_colours);
+      to_prune = prune_collapse_layer_x(iteration, graphs);
     } else if (pruning == PruningOptions::COLLAPSE_LAYER_Y) {
-      to_prune = prune_collapse_layer_y(iteration, graphs, cur_colours);
+      to_prune = prune_collapse_layer_y(iteration, graphs);
+    } else if (pruning == PruningOptions::COLLAPSE_LAYER_F) {
+      to_prune = prune_collapse_layer_f(iteration, graphs);
+    } else if (pruning == PruningOptions::COLLAPSE_LAYER_YF) {
+      to_prune = prune_collapse_layer_y(iteration, graphs);
+      std::set<int> to_prune_f = prune_collapse_layer_f(iteration, graphs);
+      to_prune.insert(to_prune_f.begin(), to_prune_f.end());
     } else {
       to_prune = std::set<int>();
       pruned = false;
@@ -72,8 +78,7 @@ namespace feature_generation {
   }
 
   std::set<int> Features::prune_collapse_layer_x(int iteration,
-                                                 const std::vector<graph::Graph> &graphs,
-                                                 std::vector<std::vector<int>> &cur_colours) {
+                                                 const std::vector<graph::Graph> &graphs) {
     int original_iterations = iterations;
     iterations = iteration;
     collecting = false;
@@ -112,8 +117,7 @@ namespace feature_generation {
   }
 
   std::set<int> Features::prune_collapse_layer_y(int iteration,
-                                                 const std::vector<graph::Graph> &graphs,
-                                                 std::vector<std::vector<int>> &cur_colours) {
+                                                 const std::vector<graph::Graph> &graphs) {
     int original_iterations = iterations;
     iterations = iteration;
     collecting = false;
@@ -128,4 +132,33 @@ namespace feature_generation {
 
     return to_prune;
   }
+
+  std::set<int> Features::prune_collapse_layer_f(int iteration,
+                                                 const std::vector<graph::Graph> &graphs) {
+    int original_iterations = iterations;
+    iterations = iteration;
+    collecting = false;
+    collected = true;
+
+    std::set<int> to_prune;
+    std::vector<Embedding> X = embed_graphs(graphs);
+    int N = (int)X.size();
+    int D = (int)X.at(0).size();
+    int one_percent = N / 100;
+    for (int i = 0; i < D; i++) {
+      int count = 0;
+      for (int j = 0; j < N; j++) {
+        count += X.at(j).at(i);
+      }
+      if (count <= one_percent) {
+        to_prune.insert(i);
+      }
+    }
+
+    collecting = true;
+    collected = false;
+    iterations = original_iterations;
+
+    return to_prune;
+  }
 }  // namespace feature_generation
diff --git a/src/feature_generation/pruning_options.cpp b/src/feature_generation/pruning_options.cpp
@@ -7,14 +7,18 @@ namespace feature_generation {
   const std::string PruningOptions::COLLAPSE_LAYER = "collapse-layer";
   const std::string PruningOptions::COLLAPSE_LAYER_X = "collapse-layer-x";
   const std::string PruningOptions::COLLAPSE_LAYER_Y = "collapse-layer-y";
+  const std::string PruningOptions::COLLAPSE_LAYER_F = "collapse-layer-f";
+  const std::string PruningOptions::COLLAPSE_LAYER_YF = "collapse-layer-yf";
   const std::vector<std::string> PruningOptions::get_all() {
     return {
         NONE,
-        COLLAPSE_ALL,
+        // COLLAPSE_ALL,
         COLLAPSE_ALL_X,
-        COLLAPSE_LAYER,
+        // COLLAPSE_LAYER,
         COLLAPSE_LAYER_X,
         COLLAPSE_LAYER_Y,
+        COLLAPSE_LAYER_F,
+        COLLAPSE_LAYER_YF,
     };
   }
 }  // namespace feature_generation