-
Notifications
You must be signed in to change notification settings - Fork 40
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
support older matlab versions (avoid using matlab arithmetic expands)
- Loading branch information
1 parent
9bc9c3d
commit 3521b31
Showing
6 changed files
with
206 additions
and
10 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,92 @@ | ||
| Dataset info - test: 10000, train: 60000, first sample size:=28 28, var=6352.04, min=0.000000, max=255.000000 | ||
| Verifying backProp.. | ||
| Checking layer 1 - input | ||
| Checking layer 2 - conv {Operation terminated by user during <a href="matlab:matlab.internal.language.introspective.errorDocCallback('feedForward', 'D:\mdCNN_git\mdCNN\feedForward.m', 55)" style="font-weight:bold">feedForward</a> (<a href="matlab: opentoline('D:\mdCNN_git\mdCNN\feedForward.m',55,0)">line 55</a>) | ||
|
|
||
|
|
||
| In <a href="matlab:matlab.internal.language.introspective.errorDocCallback('verifyBackProp', 'D:\mdCNN_git\mdCNN\verifyBackProp.m', 168)" style="font-weight:bold">verifyBackProp</a> (<a href="matlab: opentoline('D:\mdCNN_git\mdCNN\verifyBackProp.m',168,0)">line 168</a>) | ||
| netPdW = feedForward(netPdW, input, 0); | ||
|
|
||
| In <a href="matlab:matlab.internal.language.introspective.errorDocCallback('Train', 'D:\mdCNN_git\Training\Train.m', 39)" style="font-weight:bold">Train</a> (<a href="matlab: opentoline('D:\mdCNN_git\Training\Train.m',39,0)">line 39</a>) | ||
| verifyBackProp(net); | ||
| } | ||
| demoMnist | ||
| multi dimentional CNN , Hagay Garty 2016 | [email protected] | ||
| Initializing network.. | ||
| Initializing layer 1 - input | ||
| Initializing layer 2 - conv | ||
| Initializing layer 3 - conv | ||
| [Warning: Layer 3 input plus pad is 28 28 1 , not a power of 2. May reduce speed] | ||
| [> In <a href="matlab:matlab.internal.language.introspective.errorDocCallback('initNetWeight', 'D:\mdCNN_git\mdCNN\initNetWeight.m', 204)" style="font-weight:bold">initNetWeight</a> (<a href="matlab: opentoline('D:\mdCNN_git\mdCNN\initNetWeight.m',204,0)">line 204</a>) | ||
| In <a href="matlab:matlab.internal.language.introspective.errorDocCallback('CreateNet', 'D:\mdCNN_git\mdCNN\CreateNet.m', 32)" style="font-weight:bold">CreateNet</a> (<a href="matlab: opentoline('D:\mdCNN_git\mdCNN\CreateNet.m',32,0)">line 32</a>) | ||
| In <a href="matlab:matlab.internal.language.introspective.errorDocCallback('demoMnist', 'D:\mdCNN_git\Demo\MNIST\demoMnist.m', 5)" style="font-weight:bold">demoMnist</a> (<a href="matlab: opentoline('D:\mdCNN_git\Demo\MNIST\demoMnist.m',5,0)">line 5</a>)] | ||
| Initializing layer 4 - batchNorm | ||
| Initializing layer 5 - fc | ||
| Initializing layer 6 - fc | ||
| Initializing layer 7 - softmax | ||
| Initializing layer 8 - output | ||
| <strong>trainLoopCount</strong><strong> testImageNum</strong><strong> batchNum</strong><strong> ni_initial</strong><strong> ni_final</strong><strong> noImprovementTh</strong><strong> momentum</strong><strong> constInitWeight</strong><strong> lambda</strong><strong> testOnData</strong><strong> addBackround</strong><strong> testOnNull</strong><strong> augmentImage</strong><strong> augmentParams</strong><strong> centralizeImage</strong><strong> cropImage</strong><strong> flipImage</strong><strong> useRandomPatch</strong><strong> testNumPatches</strong><strong> selevtivePatchVarTh</strong><strong> testOnMiddlePatchOnly</strong><strong> normalizeNetworkInput</strong><strong> randomizeTrainingSamples</strong> | ||
| <strong>______________</strong> <strong>____________</strong> <strong>________</strong> <strong>__________</strong> <strong>________</strong> <strong>_______________</strong> <strong>________</strong> <strong>_______________</strong> <strong>______</strong> <strong>__________</strong> <strong>____________</strong> <strong>__________</strong> <strong>____________</strong> <strong>_____________</strong> <strong>_______________</strong> <strong>_________</strong> <strong>_________</strong> <strong>______________</strong> <strong>______________</strong> <strong>___________________</strong> <strong>_____________________</strong> <strong>_____________________</strong> <strong>________________________</strong> | ||
|
|
||
| 2048 1024 16 0.1 0.0005 8 0 NaN 0 0 0 0 0 [1x1 struct] 0 0 0 0 1 0 0 1 0 | ||
|
|
||
| <strong>storeMinLossNet</strong><strong> verifyBP</strong><strong> iter</strong><strong> samplesLearned</strong><strong> maxsucessRate</strong><strong> noImprovementCount</strong><strong> minLoss</strong><strong> improvementRefLoss</strong><strong> endSeed </strong> | ||
| <strong>_______________</strong> <strong>________</strong> <strong>____</strong> <strong>______________</strong> <strong>_____________</strong> <strong>__________________</strong> <strong>_______</strong> <strong>__________________</strong> <strong>____________</strong> | ||
|
|
||
| 0 1 0 0 0 0 Inf Inf [1x1 struct] | ||
|
|
||
| Layer 1: Activation=Unit, dActivation=dUnit | ||
| <strong>type </strong><strong> sizeFm </strong><strong> numFm</strong><strong> numWeights</strong><strong> Activation </strong><strong> dActivation </strong><strong> sizeOut </strong><strong> dropOut</strong> | ||
| <strong>_____</strong> <strong>______________</strong> <strong>_____</strong> <strong>__________</strong> <strong>_____________________</strong> <strong>_____________________</strong> <strong>____________</strong> <strong>_______</strong> | ||
|
|
||
| input 28 28 1 1 0 [1x1 function_handle] [1x1 function_handle] [1x4 double] 1 | ||
|
|
||
| Layer 2: Activation=Sigmoid, dActivation=dSigmoid | ||
| <strong>type</strong><strong> numFm</strong><strong> kernel </strong><strong> pad </strong><strong> dropOut</strong><strong> Activation </strong><strong> dActivation </strong><strong> inputDim</strong><strong> stride </strong><strong> pooling </strong><strong> sizeFm </strong><strong> numWeights</strong><strong> indexesStride </strong><strong> sizeOut </strong> | ||
| <strong>____</strong> <strong>_____</strong> <strong>___________</strong> <strong>___________</strong> <strong>_______</strong> <strong>_____________________</strong> <strong>_____________________</strong> <strong>________</strong> <strong>___________</strong> <strong>___________</strong> <strong>______________</strong> <strong>__________</strong> <strong>_____________________________________</strong> <strong>____________</strong> | ||
|
|
||
| conv 12 5 5 1 2 2 0 1 [1x1 function_handle] [1x1 function_handle] 2 1 1 1 1 1 1 28 28 1 312 [1x28 double] [1x28 double] [1] [1x4 double] | ||
|
|
||
| Layer 3: Activation=Sigmoid, dActivation=dSigmoid | ||
| <strong>type</strong><strong> numFm</strong><strong> kernel </strong><strong> dropOut</strong><strong> Activation </strong><strong> dActivation </strong><strong> inputDim</strong><strong> stride </strong><strong> pad </strong><strong> pooling </strong><strong> sizeFm </strong><strong> numWeights</strong><strong> indexesStride </strong><strong> sizeOut </strong> | ||
| <strong>____</strong> <strong>_____</strong> <strong>______________</strong> <strong>_______</strong> <strong>_____________________</strong> <strong>_____________________</strong> <strong>________</strong> <strong>___________</strong> <strong>___________</strong> <strong>___________</strong> <strong>______________</strong> <strong>__________</strong> <strong>_____________________________________</strong> <strong>____________</strong> | ||
|
|
||
| conv 24 13 13 1 1 [1x1 function_handle] [1x1 function_handle] 2 1 1 1 0 0 0 1 1 1 16 16 1 48696 [1x16 double] [1x16 double] [1] [1x4 double] | ||
|
|
||
| Layer 4: Activation=Unit, dActivation=dUnit | ||
| <strong> type </strong><strong> numFm</strong><strong> EPS </strong><strong> niFactor</strong><strong> Activation </strong><strong> dActivation </strong><strong> initGamma</strong><strong> initBeta</strong><strong> numWeights</strong><strong> alpha </strong><strong> dropOut</strong><strong> sizeFm </strong><strong> sizeOut </strong> | ||
| <strong>_________</strong> <strong>_____</strong> <strong>_____</strong> <strong>________</strong> <strong>_____________________</strong> <strong>_____________________</strong> <strong>_________</strong> <strong>________</strong> <strong>__________</strong> <strong>_______</strong> <strong>_______</strong> <strong>______________</strong> <strong>____________</strong> | ||
|
|
||
| batchNorm 24 1e-05 1 [1x1 function_handle] [1x1 function_handle] 1 0 12288 0.03125 1 16 16 1 [1x4 double] | ||
|
|
||
| Layer 5: Activation=Sigmoid, dActivation=dSigmoid | ||
| <strong>type</strong><strong> numFm</strong><strong> dropOut</strong><strong> Activation </strong><strong> dActivation </strong><strong> sizeFm</strong><strong> numWeights</strong><strong> sizeOut </strong> | ||
| <strong>____</strong> <strong>_____</strong> <strong>_______</strong> <strong>_____________________</strong> <strong>_____________________</strong> <strong>______</strong> <strong>__________</strong> <strong>________</strong> | ||
|
|
||
| fc 128 0.8 [1x1 function_handle] [1x1 function_handle] 1 7.8656e+05 1 128 | ||
|
|
||
| Layer 6: Activation=Sigmoid, dActivation=dSigmoid | ||
| <strong>type</strong><strong> numFm</strong><strong> dropOut</strong><strong> Activation </strong><strong> dActivation </strong><strong> sizeFm</strong><strong> numWeights</strong><strong> sizeOut</strong> | ||
| <strong>____</strong> <strong>_____</strong> <strong>_______</strong> <strong>_____________________</strong> <strong>_____________________</strong> <strong>______</strong> <strong>__________</strong> <strong>_______</strong> | ||
|
|
||
| fc 10 1 [1x1 function_handle] [1x1 function_handle] 1 1290 1 10 | ||
|
|
||
| Layer 7: Activation=Unit, dActivation=dUnit | ||
| <strong> type </strong><strong> numFm</strong><strong> Activation </strong><strong> dActivation </strong><strong> dropOut</strong><strong> sizeFm</strong><strong> numWeights</strong><strong> sizeOut</strong> | ||
| <strong>_______</strong> <strong>_____</strong> <strong>_____________________</strong> <strong>_____________________</strong> <strong>_______</strong> <strong>______</strong> <strong>__________</strong> <strong>_______</strong> | ||
|
|
||
| softmax 10 [1x1 function_handle] [1x1 function_handle] 1 1 0 1 10 | ||
|
|
||
| Layer 8: Activation=Unit, dActivation=dUnit | ||
| <strong> type </strong><strong> lossFunc </strong><strong> costFunc </strong><strong> sizeFm</strong><strong> numFm</strong><strong> sizeOut</strong><strong> Activation </strong><strong> dActivation </strong><strong> numWeights</strong> | ||
| <strong>______</strong> <strong>_____________________</strong> <strong>_____________________</strong> <strong>______</strong> <strong>_____</strong> <strong>_______</strong> <strong>_____________________</strong> <strong>_____________________</strong> <strong>__________</strong> | ||
|
|
||
| output [1x1 function_handle] [1x1 function_handle] 1 10 1 10 [1x1 function_handle] [1x1 function_handle] 0 | ||
|
|
||
| Network properties: | ||
|
|
||
| <strong>skipLastLayerErrorCalc</strong><strong> numLayers</strong><strong> version</strong><strong> sources </strong><strong> numWeights</strong><strong> numOutputs</strong> | ||
| <strong>______________________</strong> <strong>_________</strong> <strong>_______</strong> <strong>____________</strong> <strong>__________</strong> <strong>__________</strong> | ||
|
|
||
| 1 8 2.3 [4x1 struct] 8.4915e+05 10 | ||
|
|
Oops, something went wrong.