weiyuqing/ECG_by_Diffenert_CNN_Test
1
0 Fork 0
mirror of https://github.com/This-weiyuqing/ECG_by_Diffenert_CNN_Test.git synced 2026-10-18 03:13:39 +08:00
Code Issues Packages Projects Releases Wiki Activity
40 Commits 1 Branch 0 Tags
95762f507ea841582b418e6e08e46a9d3bc11aa5
Commit Graph
8 Commits
Author SHA1 Message Date
weiyuqing af187d92e9 may be last version 
By completing the thesis, we have corrected the deficiencies and fixed some bugs.
 2023-04-18 18:43:14 +08:00
weiyuqing 5b25290ebb first Degradation was 21 kernel size of conv 2023-03-24 16:22:14 +08:00
weiyuqing ba1ccb8a83 take 51 conv kernel size test 
this test prove in 101 kernel size the Degradation has occurred
 2023-03-24 15:59:04 +08:00
weiyuqing 9e65327579 201 101 conv kernel size test 
101<201and19  did this Degradation?
 2023-03-24 15:49:35 +08:00
weiyuqing 2adac589c1 conv to still to nineteen 
Until the convolution kernel size is 19 still does not reduce the degradation phenomenon, based on ECG features alone, will occur after a certain convolution kernel poorer classification performance phenomenon.
 2023-03-24 15:23:57 +08:00
weiyuqing 4924644e83 fix bug :one layer beter that two and other 
这个bug是因为全连接层选择了错误的函数。Relu回导致如此，而更换为softmax可以解决此问题。
this bug because of fc used false function :Relu will take this question 👎  and softmax fix this question 👍
 2023-03-23 21:24:08 +08:00
weiyuqing e32e008fb6 3.5.7 conv 2023-03-23 14:22:26 +08:00
weiyuqing e0a785b422 rename layer .h5 file 2023-03-23 13:29:11 +08:00