简单的神经网络模型 - 学习与保存异或逻辑关系 - Keras实现

import numpy as np
from keras.models import Sequential,model_from_json,load_model
from keras.layers import Dense,Activation
from keras.optimizers import SGD

class MNN(object):

	def __init__(self,x_train=None,y_train=None,numN=20,inputDim=2,lr=0.01,epochs=10000):
		
		self.x_train = np.array([[0,0],[0,1],[1,1],[1,0]]) if x_train == None else np.array(x_train)
		self.y_train = np.array([0,1,0,1]).T if y_train == None else np.array(y_train).T
		self.numN = numN
		self.inputDim = inputDim
		self.lr = lr
		self.epochs = epochs

	def myModel(self):

		model = Sequential()
		model.add(Dense(self.numN,input_dim=self.inputDim))
		model.add(Activation('tanh'))
		model.add(Dense(1))
		model.add(Activation('sigmoid'))

		return model

	def train(self):
		model = self.myModel()

		sgd = SGD(lr=self.lr)
		model.compile(loss="binary_crossentropy",optimizer=sgd)

		model.fit(self.x_train,self.y_train,epochs=self.epochs,verbose=0)

		return model

	def train_and_save(self,path='my_model.h5'):
		model = self.train()

		model.save(path)
		del model

	def pred(self,data,path='my_model.h5'):

		model = load_model(path)
		
		result = model.predict(np.array([data]))
		if result > 0.5:
			return 1
		else:
			return 0


				

if __name__ == "__main__":
	mnn = MNN()
	print(mnn.pred([1,1]))
	print(mnn.pred([1,0]))
	print(mnn.pred([0,1]))
	print(mnn.pred([0,0]))