Execution Time0.33s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: PR-4623-x86_64-mac1014-clang100-opt (macphsft17.dyndns.cern.ch) on 2019-11-14 10:11:32

Test Timing: Passed
Processors1

Show Command Line
Display graphs:

Test output
Testing Backpropagation:
DEEP NEURAL NETWORK:   Depth = 3  Input = ( 1, 10, 4 )  Batch size = 10  Loss function = R
	Layer 0	 DENSE Layer: 	 ( Input =     4 , Width =     2 ) 	Output = (  1 ,    10 ,     2 ) 	 Activation Function = Identity
	Layer 1	 BATCH NORM Layer: 	 ( Input =     2 ) 
	Layer 2	 DENSE Layer: 	 ( Input =     2 , Width =     1 ) 	Output = (  1 ,    10 ,     1 ) 	 Activation Function = Identity
input 

10x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.9989     -0.4348      0.7818    -0.03005 
   1 |     0.8243    -0.05672     -0.9009     -0.0747 
   2 |   0.007912     -0.4108       1.391     -0.9851 
   3 |   -0.04894      -1.443      -1.061      -1.388 
   4 |     0.7674      -0.736      0.5797     -0.3821 
   5 |      2.061      -1.235       1.165     -0.4542 
   6 |    -0.1348     -0.4996     -0.1824       1.844 
   7 |    -0.2428       1.997    0.004806     -0.4222 
   8 |      1.541     0.09474       1.525       1.217 
   9 |    -0.1363     -0.1992     -0.2938     -0.1184 

 training batch 1 mu var0-0.0817923
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.4066      0.3507 
   1 |     0.4697     -0.9611 
   2 |    -0.4395      0.7171 
   3 |      2.232      -2.589 
   4 |     0.1127      -0.181 
   5 |    -0.1716    -0.09331 
   6 |     0.3033     -0.0499 
   7 |     -1.628        1.52 
   8 |     -1.721       1.832 
   9 |     0.4308      -0.473 

output BN 
output DL feature 0 mean -0.0817923	output DL std 1.12664
output DL feature 1 mean 0.00726289	output DL std 1.25599
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.3039      0.2882 
   1 |     0.5159     -0.8127 
   2 |    -0.3347      0.5957 
   3 |      2.165      -2.179 
   4 |     0.1819      -0.158 
   5 |   -0.08407     -0.0844 
   6 |     0.3603    -0.04797 
   7 |     -1.446        1.27 
   8 |     -1.534       1.531 
   9 |     0.4796     -0.4031 

output BN feature 0 mean -3.33067e-17	output BN std 1.05405
output BN feature 1 mean 2.22045e-17	output BN std 1.05406
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    0.07236     0.03787    -0.05939    -0.06988 
   1 |    -0.1898     -0.9295      -1.109     -0.7049 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    -0.2563     -0.8785     -0.6873     -0.1602 
   1 |     -0.036      0.8152      0.9589       0.286 

 training batch 2 mu var0-0.0817895
compute loss for weight  -0.256311  -0.256321 result 2.83654
 training batch 3 mu var0-0.0817923
compute loss for weight  -0.256331  -0.256321 result 2.83654
 training batch 4 mu var0-0.0817916
compute loss for weight  -0.256316  -0.256321 result 2.83654
 training batch 5 mu var0-0.0817923
compute loss for weight  -0.256326  -0.256321 result 2.83654
   --dy = 0.0723643 dy_ref = 0.0723643
 training batch 6 mu var0-0.0817928
compute loss for weight  -0.878452  -0.878462 result 2.83654
 training batch 7 mu var0-0.0817923
compute loss for weight  -0.878472  -0.878462 result 2.83654
 training batch 8 mu var0-0.0817925
compute loss for weight  -0.878457  -0.878462 result 2.83654
 training batch 9 mu var0-0.0817923
compute loss for weight  -0.878467  -0.878462 result 2.83654
   --dy = 0.0378743 dy_ref = 0.0378743
 training batch 10 mu var0-0.081792
compute loss for weight  -0.687272  -0.687282 result 2.83654
 training batch 11 mu var0-0.0817923
compute loss for weight  -0.687292  -0.687282 result 2.83654
 training batch 12 mu var0-0.0817922
compute loss for weight  -0.687277  -0.687282 result 2.83654
 training batch 13 mu var0-0.0817923
compute loss for weight  -0.687287  -0.687282 result 2.83654
   --dy = -0.0593937 dy_ref = -0.0593937
 training batch 14 mu var0-0.0817924
compute loss for weight  -0.160221  -0.160231 result 2.83654
 training batch 15 mu var0-0.0817923
compute loss for weight  -0.160241  -0.160231 result 2.83654
 training batch 16 mu var0-0.0817924
compute loss for weight  -0.160226  -0.160231 result 2.83654
 training batch 17 mu var0-0.0817923
compute loss for weight  -0.160236  -0.160231 result 2.83654
   --dy = -0.0698751 dy_ref = -0.0698751
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      2.248       3.425 

weights for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |          1           1 

 training batch 18 mu var0-0.0817923
compute loss for weight  1.00001  1 result 2.83656
 training batch 19 mu var0-0.0817923
compute loss for weight  0.99999  1 result 2.83652
 training batch 20 mu var0-0.0817923
compute loss for weight  1.00001  1 result 2.83655
 training batch 21 mu var0-0.0817923
compute loss for weight  0.999995  1 result 2.83653
   --dy = 2.24767 dy_ref = 2.24767
 training batch 22 mu var0-0.0817923
compute loss for weight  1.00001  1 result 2.83658
 training batch 23 mu var0-0.0817923
compute loss for weight  0.99999  1 result 2.83651
 training batch 24 mu var0-0.0817923
compute loss for weight  1.00001  1 result 2.83656
 training batch 25 mu var0-0.0817923
compute loss for weight  0.999995  1 result 2.83652
   --dy = 3.42541 dy_ref = 3.42541
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 | -1.527e-16  -8.327e-17 

weights for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |          0           0 

 training batch 26 mu var0-0.0817923
compute loss for weight  1e-05  0 result 2.83654
 training batch 27 mu var0-0.0817923
compute loss for weight  -1e-05  0 result 2.83654
 training batch 28 mu var0-0.0817923
compute loss for weight  5e-06  0 result 2.83654
 training batch 29 mu var0-0.0817923
compute loss for weight  -5e-06  0 result 2.83654
   --dy = -5.92119e-11 dy_ref = -1.52656e-16
 training batch 30 mu var0-0.0817923
compute loss for weight  1e-05  0 result 2.83654
 training batch 31 mu var0-0.0817923
compute loss for weight  -1e-05  0 result 2.83654
 training batch 32 mu var0-0.0817923
compute loss for weight  5e-06  0 result 2.83654
 training batch 33 mu var0-0.0817923
compute loss for weight  -5e-06  0 result 2.83654
   --dy = 0 dy_ref = -8.32667e-17
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     -3.349        3.36 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.6712        1.02 

 training batch 34 mu var0-0.0817923
compute loss for weight  -0.671215  -0.671225 result 2.83651
 training batch 35 mu var0-0.0817923
compute loss for weight  -0.671235  -0.671225 result 2.83657
 training batch 36 mu var0-0.0817923
compute loss for weight  -0.67122  -0.671225 result 2.83652
 training batch 37 mu var0-0.0817923
compute loss for weight  -0.67123  -0.671225 result 2.83656
   --dy = -3.34861 dy_ref = -3.34861
 training batch 38 mu var0-0.0817923
compute loss for weight  1.01954  1.01953 result 2.83657
 training batch 39 mu var0-0.0817923
compute loss for weight  1.01952  1.01953 result 2.83651
 training batch 40 mu var0-0.0817923
compute loss for weight  1.01954  1.01953 result 2.83656
 training batch 41 mu var0-0.0817923
compute loss for weight  1.01953  1.01953 result 2.83652
   --dy = 3.35978 dy_ref = 3.35978
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][33m1.7206e-09[NON-XML-CHAR-0x1B][39m