add prediction

YJ-Keras/pred.py

@@ -25,6 +25,8 @@
 
 parser=argparse.ArgumentParser()
 parser.add_argument("--rat",type=float,default=0.6,help='ratio for weak qg batch')
+parser.add_argument("--epoch",type=int,default=10,help='epoch')
+parser.add_argument("--load",type=str,default="weakdijet_0.8",help='load name')
 parser.add_argument("--save",type=str,default=1,help='rch')
 args=parser.parse_args()
 
@@ -47,8 +49,8 @@
 #              optimizer=keras.optimizers.Adadelta(),
 #              metrics=['accuracy'])
 #model=keras.models.load_model('save/fullydijetsame_10')
-savename=args.save
-model=keras.models.load_model(savename)
+savename="save/"+args.load
+model=keras.models.load_model(savename+"/_"+str(args.epoch))
 
 #train=wkiter(["root/cutb/q"+str(int(args.rat*100))+"img.root","root/cutb/g"+str(int(args.rat*100))+"img.root"],batch_size=batch_size,end=1,istrain=1,friend=0)
 test=wkiter("root",friend=20,begin=5./7.,end=1.,batch_size=batch_size)
@@ -74,13 +76,13 @@
 plt.hist(q,bins=50,weights=np.ones_like(q),histtype='step',alpha=0.7,label='quark')
 plt.hist(g,bins=50,weights=np.ones_like(g),histtype='step',alpha=0.7,label='gluon')
 plt.legend(loc="upper center")
-plt.savefig(savename+"like.png")
+plt.savefig(savename+"/like"+str(args.epoch)+".png")
 t_fpr,t_tpr,_=roc_curve(x,y)
 t_fnr=1-t_fpr
 train_auc=np.around(auc(t_fpr,t_tpr),4)
 plt.figure(2)
 plt.plot(t_tpr,t_fnr,alpha=0.5,label="AUC={}".format(train_auc),lw=2)
 plt.legend(loc='lower left')
-plt.savefig(savename+"roc.png")
+plt.savefig(savename+"/roc"+str(args.epoch)+".png")
 #print(b,c)
 

YJ-Keras/symbols/asvgg.py

@@ -0,0 +1,83 @@
+from keras.models import Sequential
+from keras.layers.core import Flatten, Dense, Dropout
+from keras.layers.convolutional import Convolution2D, MaxPooling2D, ZeroPadding2D
+from keras.optimizers import SGD
+import cv2, numpy as np
+
+def get_symbol(input_shape,num_classes,weights_path=None):
+    model = Sequential()
+    model.add(ZeroPadding2D((1,1),input_shape=input_shape))
+    model.add(Convolution2D(64, 3, 1, activation='relu'))
+    model.add(Convolution2D(64, 1, 3, activation='relu'))
+    model.add(ZeroPadding2D((1,1)))
+    model.add(Convolution2D(64, 3, 1, activation='relu'))
+    model.add(Convolution2D(64, 1, 3, activation='relu'))
+    model.add(MaxPooling2D((2,2), strides=(2,2)))
+
+    model.add(ZeroPadding2D((1,1)))
+    model.add(Convolution2D(128, 3, 1, activation='relu'))
+    model.add(Convolution2D(128, 1, 3, activation='relu'))
+    model.add(ZeroPadding2D((1,1)))
+    model.add(Convolution2D(128, 3, 1, activation='relu'))
+    model.add(Convolution2D(128, 1, 3, activation='relu'))
+    model.add(MaxPooling2D((2,2), strides=(2,2)))
+
+    model.add(ZeroPadding2D((1,1)))
+    model.add(Convolution2D(256, 3, 1, activation='relu'))
+    model.add(Convolution2D(256, 1, 3, activation='relu'))
+    model.add(ZeroPadding2D((1,1)))
+    model.add(Convolution2D(256, 3, 1, activation='relu'))
+    model.add(Convolution2D(256, 1, 3, activation='relu'))
+    model.add(ZeroPadding2D((1,1)))
+    model.add(Convolution2D(256, 3, 1, activation='relu'))
+    model.add(Convolution2D(256, 1, 3, activation='relu'))
+    model.add(MaxPooling2D((2,2), strides=(2,2)))
+    """
+    model.add(ZeroPadding2D((1,1)))
+    model.add(Convolution2D(512, 3, 1, activation='relu'))
+    model.add(Convolution2D(512, 1, 3, activation='relu'))
+    model.add(ZeroPadding2D((1,1)))
+    model.add(Convolution2D(512, 3, 1, activation='relu'))
+    model.add(Convolution2D(512, 1, 3, activation='relu'))
+    model.add(ZeroPadding2D((1,1)))
+    model.add(Convolution2D(512, 3, 1, activation='relu'))
+    model.add(Convolution2D(512, 1, 3, activation='relu'))
+    model.add(MaxPooling2D((2,2), strides=(2,2)))
+
+    model.add(ZeroPadding2D((1,1)))
+    model.add(Convolution2D(512, 3, 1, activation='relu'))
+    model.add(Convolution2D(512, 1, 3, activation='relu'))
+    model.add(ZeroPadding2D((1,1)))
+    model.add(Convolution2D(512, 3, 1, activation='relu'))
+    model.add(Convolution2D(512, 1, 3, activation='relu'))
+    model.add(ZeroPadding2D((1,1)))
+    model.add(Convolution2D(512, 3, 1, activation='relu'))
+    model.add(Convolution2D(512, 1, 3, activation='relu'))
+    model.add(MaxPooling2D((2,2), strides=(2,2)))
+    """
+    model.add(Flatten())
+    model.add(Dense(4096, activation='relu'))
+    model.add(Dropout(0.5))
+    model.add(Dense(4096, activation='relu'))
+    model.add(Dropout(0.5))
+    model.add(Dense(num_classes, activation='softmax'))
+
+    if weights_path:
+        model.load_weights(weights_path)
+
+    return model
+
+if __name__ == "__main__":
+    im = cv2.resize(cv2.imread('cat.jpg'), (224, 224)).astype(np.float32)
+    im[:,:,0] -= 103.939
+    im[:,:,1] -= 116.779
+    im[:,:,2] -= 123.68
+    im = im.transpose((2,0,1))
+    im = np.expand_dims(im, axis=0)
+
+    # Test pretrained model
+    model = VGG_16('vgg16_weights.h5')
+    sgd = SGD(lr=0.1, decay=1e-6, momentum=0.9, nesterov=True)
+    model.compile(optimizer=sgd, loss='categorical_crossentropy')
+    out = model.predict(im)
+    print np.argmax(out)

YJ-Keras/weak.py

@@ -17,6 +17,7 @@
 import tensorflow as tf
 from keras.backend.tensorflow_backend import set_session
 from importlib import import_module
+import os
 config =tf.ConfigProto()
 config.gpu_options.per_process_gpu_memory_fraction=0.4
 set_session(tf.Session(config=config))
@@ -27,6 +28,7 @@
 
 parser=argparse.ArgumentParser()
 parser.add_argument("--rat",type=float,default=0.6,help='ratio for weak qg batch')
+parser.add_argument("--save",type=str,default="weakdijet_",help='save name')
 args=parser.parse_args()
 
 # input image dimensions
@@ -44,45 +46,27 @@
 #x_test = x_test.reshape(x_test.shape[0], img_rows, img_cols, 1)
 input_shape = (3,img_rows, img_cols)
 
-#x_train = x_train.astype('float32')
-#x_test = x_test.astype('float32')
-#x_train /= 255
-#x_test /= 255
-#print('x_train shape:', x_train.shape)
-#print(x_train.shape[0], 'train samples')
-#print(x_test.shape[0], 'test samples')
 
-# convert class vectors to binary class matrices
-#y_train = keras.utils.to_categorical(y_train, num_classes)
-#y_test = keras.utils.to_categorical(y_test, num_classes)
-
-"""model = Sequential()
-model.add(Conv2D(32, kernel_size=(3, 3),
-                 activation='relu',
-                 input_shape=input_shape))
-model.add(Conv2D(64, (3, 3), activation='relu'))
-model.add(MaxPooling2D(pool_size=(2, 2)))
-model.add(Dropout(0.25))
-model.add(Flatten())
-model.add(Dense(128, activation='relu'))
-model.add(Dropout(0.5))
-model.add(Dense(num_classes, activation='softmax'))"""
-
-net=import_module('symbols.'+"vgg")
+net=import_module('symbols.'+"asvgg")
 model=net.get_symbol(input_shape,num_classes)
 model.compile(loss=keras.losses.categorical_crossentropy,
-              optimizer=keras.optimizers.Adadelta(),
+              optimizer=keras.optimizers.SGD(),
               metrics=['accuracy'])
 
 #ab=int(len(x_train)/10)
 
-train=wkiter(["root/cutb/q"+str(int(args.rat*100))+"img.root","root/cutb/g"+str(int(args.rat*100))+"img.root"],batch_size=batch_size,end=1,istrain=1,friend=0)
-test=wkiter("root",friend=20,begin=5./7.,end=1.,batch_size=batch_size)
+train=wkiter(["root/cutb/q"+str(int(args.rat*100))+"img.root","root/cutb/g"+str(int(args.rat*100))+"img.root"],batch_size=batch_size,end=0.01,istrain=1,friend=0)
+test=wkiter("root",friend=20,begin=5./7.,end=5.1/7.,batch_size=batch_size)
+savename='save/'+str(args.save)+str(args.rat)
+os.system("mkdir "+savename)
 print ("train",train.totalnum(),"eval",test.totalnum())
+#logger=keras.callbacks.CSVLogger(savename+'/log.log',append=True)
+logger=keras.callbacks.TensorBoard(log_dir=savename+'/logs',histogram_freq=0, write_graph=True , write_images=True, batch_size=batch_size)
+
 for i in range(epochs):
 	print("epoch",i)
-	checkpoint=keras.callbacks.ModelCheckpoint(filepath='save/weakdijet_'+str(args.rat)+'_'+str(i),monitor='val_loss',verbose=0,save_best_only=False,mode='auto')
-	model.fit_generator(train.next(),steps_per_epoch=train.totalnum(),validation_data=test.next(),validation_steps=test.totalnum(),epochs=1,verbose=1,callbacks=[checkpoint])
+	checkpoint=keras.callbacks.ModelCheckpoint(filepath=savename+'/_'+str(i),monitor='val_loss',verbose=0,save_best_only=False,mode='auto')
+	model.fit_generator(train.next(),steps_per_epoch=train.totalnum(),validation_data=test.next(),validation_steps=test.totalnum(),epochs=1,verbose=1,callbacks=[logger,checkpoint])
 	train.reset()
 	test.reset()
 	"""while True:
@@ -108,13 +92,3 @@
 
 
 
-score = model.evaluate(x_test, y_test, verbose=0)
-print('Test loss:', score[0])
-print('Test accuracy:', score[1])
-
-
-"""while True:
-	i+=1
-	a,b=train.next()
-	print(a.shape,b.shape,i)
-	if(train.endfile==1):break"""