fix drne

Author: kinkunchan

.DS_Store

@@ -151,3 +151,5 @@ cython_debug/
 semb/methods/struc2vec/pickles/
 semb/methods/struc2vec/random_walks.txt
 
+# drne intermediate files
+semb/methods/drne/eni/

.gitignore

@@ -127,7 +127,7 @@ def train(self):
                 n += 1
                 end = time.time()
                 #process = psutil.Process(os.getpid())
-                print(("epoch: {}/{}, batch: {}/{}, loss: {:.6f}, structure_loss: {:.6f}, orth_loss: {:.6f}, guilded_loss: {:.6f}, time: {:.4f}s").format(epoch, self.params.epochs_to_train, n-1, total_num, total_loss, structure_loss, orth_loss, guilded_loss, end-begin))
+                print(("epoch: {}/{}, batch: {}/{}, loss: {:.6f}, structure_loss: {:.6f}, orth_loss: {:.6f}, guilded_loss: {:.6f}, time: {:.4f}s").format(epoch, self.params['epochs_to_train'], n-1, total_num, total_loss, structure_loss, orth_loss, guilded_loss, end-begin))
                 if num % 5 == 0:
                     summary_str = self.sess.run(self.merged_summary, feed_dict=batch_data)
                     self.summary_writer.add_summary(summary_str, num)

semb/.DS_Store

@@ -5,25 +5,66 @@
 import time
 import os
 import tensorflow as tf
+import copy
+import networkx as nx
 
+folder_eni = os.path.join(os.path.dirname(__file__), "eni/")
 
 class Method(BaseMethod):
 
-    __PARAMS__ = dict(embedding_path=16, epochs_to_train=20, batch_size=16, learning_rate=0.0025,
+    __PARAMS__ = dict(embedding_size=128, epochs_to_train=20, batch_size=256, learning_rate=0.0025,
                       alpha=0.0, lamb=0.5, grad_clip=5.0, k=1, sampling_size=100,
-                      seed=1, index_from_0=True, train_device='cpu', save_path=os.getcwd())
+                      seed=1, index_from_0=True, train_device='cpu', save_path=folder_eni, save_suffix='eni')
 
     def get_id(self):
         return "drne"
 
     def train(self):
         np.random.seed(int(time.time())
                        if self.params['seed'] == -1 else self.params['seed'])
+
+        self.old_graph = copy.deepcopy(self.graph)
+        old_edges = [e for e in self.old_graph.edges()]
+        dict_node_o2n = dict()
+        list_edges = list()
+
+        count = 1
+        for edge in old_edges:
+            src = int(edge[0])
+            dst = int(edge[1])
+
+            if src not in dict_node_o2n:
+                dict_node_o2n[src] = count
+                count += 1
+            if dst not in dict_node_o2n:
+                dict_node_o2n[dst] = count
+                count += 1
+            list_edges += [(dict_node_o2n[src], dict_node_o2n[dst])]
+
+        graph = [[]]
+        for _ in range(len(dict_node_o2n)):
+            graph += [[]]
+
+        for cur_edge in list_edges:
+            src = cur_edge[0]
+            dst = cur_edge[1]
+            if dst not in graph[src]:
+                graph[src] += [dst]
+            if src not in graph[dst]:
+                graph[dst] += [src]
+
+        self.graph = graph
+
         network.sort_graph_by_degree(self.graph)
         config = tf.ConfigProto(allow_soft_placement=True)
         config.gpu_options.allow_growth = True
         with tf.Graph().as_default(), tf.Session(config=config) as sess, tf.device(self.params['train_device']):
             alg = eni(self.graph, self.params, sess)
-            print("max degree: {}".format(alg.degree_max))
+            # print("max degree: {}".format(alg.degree_max))
             alg.train()
-            self.embeddings = alg.get_embeddings()
+            reps = alg.get_embeddings()
+
+        dict_n2o = {val: key for key, val in dict_node_o2n.items()}
+        self.embeddings = dict()
+        for i in range(0, reps.shape[0]):
+            self.embeddings[dict_n2o[i+1]] = reps[i].tolist()