标签: deep-learning lstm recurrent-neural-network bayesian-networks dropout
我有一个使用keras开发的简单LSTM网络
model = Sequential() model.add(LSTM(rnn_size,input_shape=(2,w),dropout = 0.25 , recurrent_dropout=0.25)) model.add(Dense(2))
并且我想应用MC Dropout方法,如何在测试阶段启用Dropout以计算不确定性。 谢谢