update

Author: zhedongzheng

nlp-models/tfidf_logistic_test.py

@@ -0,0 +1,33 @@
+import tensorflow as tf
+import numpy as np
+from sklearn.feature_extraction.text import TfidfTransformer
+from sklearn.linear_model import LogisticRegression
+
+
+vocab_size = 20000
+
+
+if __name__ == '__main__':
+    (X_train, y_train), (X_test, y_test) = tf.contrib.keras.datasets.imdb.load_data(num_words=vocab_size)
+
+    indexed = np.concatenate([X_train, X_test])
+    DT = np.zeros((len(indexed), vocab_size))
+    i = 0
+    for indices in indexed:
+        for idx in indices:
+            DT[i, idx] += 1
+        i += 1
+    print("Document-Term matrix built ...")
+
+    model = TfidfTransformer()
+    DT = model.fit_transform(DT).toarray()
+    print("TF-IDF transform completed ...")
+    
+    X_train = DT[:len(X_train)]
+    X_test = DT[len(X_train):]
+
+    model = LogisticRegression()
+    y_pred = model.fit(X_train, y_train).predict(X_test)
+    
+    final_acc = (y_pred == y_test).mean()
+    print("final testing accuracy: %.4f" % final_acc)

nlp-models/tfidf_tsne_test.py

@@ -11,22 +11,23 @@
     vocab_size = len(word2idx)
     print("Data loaded | Vocab size:", vocab_size, '| Document size:', len(indexed))
 
-    TD = np.zeros((vocab_size, len(indexed))) # term-document matrix
-    j = 0
+    DT = np.zeros((len(indexed), vocab_size)) # document-term matrix
+    i = 0
     for indices in indexed:
         for idx in indices:
-            TD[idx, j] += 1
-        j += 1
-    print("Term-Document matrix built ...")
+            DT[i, idx] += 1
+        i += 1
+    print("Document-Term matrix built ...")
 
     model = TfidfTransformer()
-    DT = TD.T
     DT = model.fit_transform(DT).toarray()
     TD = DT.T
     print("TF-IDF transform completed ...")
 
     find_analogy('london', TD, word2idx)
     find_analogy('king', TD, word2idx)
+    find_analogy('italy', TD, word2idx)
+    find_analogy('queen', TD, word2idx)
 
     """
     model = TSNE(n_components=2, verbose=2, learning_rate=200)

tensorflow-models/cnn/concat_conv_1d_text_clf_imdb_test.py

@@ -23,5 +23,5 @@
                   val_data=(X_test,Y_test))
     pred = clf.predict(X_test)
 
-    final_acc = (np.argmax(pred,1) == np.argmax(Y_test,1)).astype(float).mean()
+    final_acc = (np.argmax(pred,1) == np.argmax(Y_test,1)).mean()
     print("final testing accuracy: %.4f" % final_acc)

tensorflow-models/cnn/conv_1d_text_clf_imdb_test.py

@@ -23,5 +23,5 @@
                   val_data=(X_test,Y_test))
     pred = clf.predict(X_test)
 
-    final_acc = (np.argmax(pred,1) == np.argmax(Y_test,1)).astype(float).mean()
+    final_acc = (np.argmax(pred,1) == np.argmax(Y_test,1)).mean()
     print("final testing accuracy: %.4f" % final_acc)

tensorflow-models/cnn/conv_2d_clf_cifar10_test.py

@@ -14,5 +14,5 @@
     log = clf.fit(X_train, Y_train, val_data=(X_test,Y_test))
     pred = clf.predict(X_test)
 
-    final_acc = (np.argmax(pred,1) == np.argmax(Y_test,1)).astype(float).mean()
+    final_acc = (np.argmax(pred,1) == np.argmax(Y_test,1)).mean()
     print("final testing accuracy: %.4f" % final_acc)

tensorflow-models/cnn/conv_2d_clf_mnist_test.py

@@ -14,5 +14,5 @@
     log = clf.fit(X_train, Y_train, val_data=(X_test,Y_test))
     pred = clf.predict(X_test)
 
-    final_acc = (np.argmax(pred,1) == np.argmax(Y_test,1)).astype(float).mean()
+    final_acc = (np.argmax(pred,1) == np.argmax(Y_test,1)).mean()
     print("final testing accuracy: %.4f" % final_acc)

tensorflow-models/cnn_rnn/conv_rnn_text_clf_imdb_test.py

@@ -22,5 +22,5 @@
     log = clf.fit(X_train, Y_train, batch_size=32, n_epoch=3, keep_prob=0.8, val_data=(X_test,Y_test))
     pred = clf.predict(X_test)
 
-    final_acc = (np.argmax(pred,1) == np.argmax(Y_test,1)).astype(float).mean()
+    final_acc = (np.argmax(pred,1) == np.argmax(Y_test,1)).mean()
     print("final testing accuracy: %.4f" % final_acc)

tensorflow-models/highway/conv_1d_hn_text_clf_imdb_test.py

@@ -23,5 +23,5 @@
                   val_data=(X_test,Y_test))
     pred = clf.predict(X_test)
 
-    final_acc = (np.argmax(pred,1) == np.argmax(Y_test,1)).astype(float).mean()
+    final_acc = (np.argmax(pred,1) == np.argmax(Y_test,1)).mean()
     print("final testing accuracy: %.4f" % final_acc)

tensorflow-models/highway/mlp_hn_clf_cifar10_test.py

@@ -15,5 +15,5 @@
     log = clf.fit(X_train, Y_train, n_epoch=30, en_exp_decay=False, val_data=(X_test,Y_test))
     pred = clf.predict(X_test)
 
-    final_acc = (np.argmax(pred,1) == np.argmax(Y_test,1)).astype(float).mean()
+    final_acc = (np.argmax(pred,1) == np.argmax(Y_test,1)).mean()
     print("final testing accuracy: %.4f" % final_acc)

tensorflow-models/highway/mlp_hn_clf_mnist_test.py

@@ -14,5 +14,5 @@
     log = clf.fit(X_train, Y_train, val_data=(X_test,Y_test))
     pred = clf.predict(X_test)
 
-    final_acc = (np.argmax(pred, 1) == np.argmax(Y_test, 1)).astype(float).mean()
+    final_acc = (np.argmax(pred, 1) == np.argmax(Y_test, 1)).mean()
     print("final testing accuracy: %.4f" % final_acc)