https://blog.youkuaiyun.com/Suodisliezh-cnhttps://blog.youkuaiyun.com/https://blog.youkuaiyun.com/Suodislie/article/details/81871557https://blog.youkuaiyun.com/Suodislie/article/details/81871557SuodislieMon, 20 Aug 2018 15:35:12 +0800https://blog.youkuaiyun.com/Suodislie/article/details/79912670https://blog.youkuaiyun.com/Suodislie/article/details/79912670SuodislieThu, 12 Apr 2018 14:17:26 +0800https://blog.youkuaiyun.com/Suodislie/article/details/79912310https://blog.youkuaiyun.com/Suodislie/article/details/79912310SuodislieThu, 12 Apr 2018 13:42:14 +0800https://blog.youkuaiyun.com/Suodislie/article/details/79607489https://blog.youkuaiyun.com/Suodislie/article/details/79607489SuodislieMon, 19 Mar 2018 09:59:02 +0800https://blog.youkuaiyun.com/Suodislie/article/details/77744668https://blog.youkuaiyun.com/Suodislie/article/details/77744668SuodislieThu, 31 Aug 2017 12:10:14 +0800https://blog.youkuaiyun.com/Suodislie/article/details/77074173https://blog.youkuaiyun.com/Suodislie/article/details/77074173SuodislieThu, 10 Aug 2017 22:07:26 +0800https://blog.youkuaiyun.com/Suodislie/article/details/74933958https://blog.youkuaiyun.com/Suodislie/article/details/74933958SuodislieMon, 10 Jul 2017 19:06:44 +0800https://blog.youkuaiyun.com/Suodislie/article/details/72823831https://blog.youkuaiyun.com/Suodislie/article/details/72823831SuodislieWed, 31 May 2017 20:55:57 +0800正则值>dropout 学习率：0.001，0.01，0.1，1，10  …….以10为阶数尝试 小数据集上合适的参数大数据集上一般不会差，可以先减少训练类别。   欠拟合： 表现形式：训练集，测试集准确率都很低 解决办法：增加网络层数，增加节点数，减少dropout值，减少L2正则值等等     过拟合： 表现形式：训练集准确率较高，测试集准确率比较低 解决]]>https://blog.youkuaiyun.com/Suodislie/article/details/72772008https://blog.youkuaiyun.com/Suodislie/article/details/72772008SuodislieFri, 26 May 2017 17:22:20 +0800https://blog.youkuaiyun.com/Suodislie/article/details/72357870https://blog.youkuaiyun.com/Suodislie/article/details/72357870SuodislieWed, 17 May 2017 10:46:29 +0800https://blog.youkuaiyun.com/Suodislie/article/details/72301559https://blog.youkuaiyun.com/Suodislie/article/details/72301559SuodislieTue, 16 May 2017 16:04:09 +0800https://blog.youkuaiyun.com/Suodislie/article/details/71440236https://blog.youkuaiyun.com/Suodislie/article/details/71440236SuodislieTue, 09 May 2017 11:04:44 +0800https://blog.youkuaiyun.com/Suodislie/article/details/70308037https://blog.youkuaiyun.com/Suodislie/article/details/70308037SuodislieFri, 21 Apr 2017 09:12:25 +0800https://blog.youkuaiyun.com/Suodislie/article/details/70228138https://blog.youkuaiyun.com/Suodislie/article/details/70228138SuodislieTue, 18 Apr 2017 14:31:44 +0800