DL之CNN：利用自定义DeepConvNet【7+1】算法对mnist数据集训练实现手写数字识别、模型评估(99.4%)

输出结果

设计思路

核心代码

network = DeepConvNet()                        

network.load_params("data_input/DeepConvNet/deep_convnet_params.pkl")  

#T1、caluculate accuracy(float64)

print("DeepConvNet【7+1】 on mnist：caluculate accuracy (float64 type) ... ")

print(network.accuracy(x_test, t_test))          #caluculate accuracy(float64)

#T2、caluculate accuracy(float16)

x_test = x_test.astype(np.float16)        

for param in network.params.values():    

   param[...] = param.astype(np.float16)

print("DeepConvNet【7+1】 on mnist：caluculate accuracy (float16 type) ... ")

print(network.accuracy(x_test, t_test))