Tensorflow2.0——过拟合问题Overfitting
Tensorflow2.0——OverfittingOverfittingregularization/weight decay动量与学习率momentum动量learning rate decayEarly StoppingdropoutStochastic Gradient DescentOverfitting解决办法:1.增加数据量2.降低模型复杂度:控制网络结构、增加正则项(惩罚项)3.dropout4.数据增强5.测试数据在过拟合前停止迭代regularization/weight