finish hw2

ldsfcj · Sep 17, 2018 · f5a8613 · f5a8613
1 parent 6e55df6
commit f5a8613
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/HW2/README.md b/HW2/README.md
@@ -37,7 +37,7 @@
 ![](https://github.com/maplezzz/ML2017S_Hung-yi-Lee_HW/blob/master/HW2/output/TrainProcess.png)  
 随着epoch-time的增多，loss不断减小，最后valid accuracy为0.852858, test accuracy为0.852343，优于Probability Generative Model这是因为这个模型下不需要假设采样数据的分布  
 
-**Neural Network** 
+**Neural Network**   
 使用keras三层的fully-connected neural network，使用的loss是binary_crossentropy，activation='sigmoid'，optimizer='adam'，前两层都是600个units，batch-size=32， epoch-times=50  
 最后在valid set上的acc=0.9084, test-set上的acc为0.8426