Logo ROOT  
Reference Guide
 
Loading...
Searching...
No Matches
TMVA_CNN_Classification.py File Reference

Detailed Description

View in nbviewer Open in SWAN
TMVA Classification Example Using a Convolutional Neural Network

This is an example of using a CNN in TMVA. We do classification using a toy image data set that is generated when running the example macro

Running with nthreads = 4
DataSetInfo : [dataset] : Added class "Signal"
: Add Tree sig_tree of type Signal with 1000 events
DataSetInfo : [dataset] : Added class "Background"
: Add Tree bkg_tree of type Background with 1000 events
Factory : Booking method: ␛[1mBDT␛[0m
:
: Rebuilding Dataset dataset
: Building event vectors for type 2 Signal
: Dataset[dataset] : create input formulas for tree sig_tree
: Using variable vars[0] from array expression vars of size 256
: Building event vectors for type 2 Background
: Dataset[dataset] : create input formulas for tree bkg_tree
: Using variable vars[0] from array expression vars of size 256
DataSetFactory : [dataset] : Number of events in input trees
:
:
: Number of training and testing events
: ---------------------------------------------------------------------------
: Signal -- training events : 800
: Signal -- testing events : 200
: Signal -- training and testing events: 1000
: Background -- training events : 800
: Background -- testing events : 200
: Background -- training and testing events: 1000
:
Factory : Booking method: ␛[1mTMVA_DNN_CPU␛[0m
:
: Parsing option string:
: ... "!H:V:ErrorStrategy=CROSSENTROPY:VarTransform=None:WeightInitialization=XAVIER:Layout=DENSE|100|RELU,BNORM,DENSE|100|RELU,BNORM,DENSE|100|RELU,BNORM,DENSE|100|RELU,DENSE|1|LINEAR:TrainingStrategy=LearningRate=1e-3,Momentum=0.9,Repetitions=1,ConvergenceSteps=5,BatchSize=100,TestRepetitions=1,WeightDecay=1e-4,Regularization=None,Optimizer=ADAM,DropConfig=0.0+0.0+0.0+0.,MaxEpochs=10:Architecture=CPU"
: The following options are set:
: - By User:
: <none>
: - Default:
: Boost_num: "0" [Number of times the classifier will be boosted]
: Parsing option string:
: ... "!H:V:ErrorStrategy=CROSSENTROPY:VarTransform=None:WeightInitialization=XAVIER:Layout=DENSE|100|RELU,BNORM,DENSE|100|RELU,BNORM,DENSE|100|RELU,BNORM,DENSE|100|RELU,DENSE|1|LINEAR:TrainingStrategy=LearningRate=1e-3,Momentum=0.9,Repetitions=1,ConvergenceSteps=5,BatchSize=100,TestRepetitions=1,WeightDecay=1e-4,Regularization=None,Optimizer=ADAM,DropConfig=0.0+0.0+0.0+0.,MaxEpochs=10:Architecture=CPU"
: The following options are set:
: - By User:
: V: "True" [Verbose output (short form of "VerbosityLevel" below - overrides the latter one)]
: VarTransform: "None" [List of variable transformations performed before training, e.g., "D_Background,P_Signal,G,N_AllClasses" for: "Decorrelation, PCA-transformation, Gaussianisation, Normalisation, each for the given class of events ('AllClasses' denotes all events of all classes, if no class indication is given, 'All' is assumed)"]
: H: "False" [Print method-specific help message]
: Layout: "DENSE|100|RELU,BNORM,DENSE|100|RELU,BNORM,DENSE|100|RELU,BNORM,DENSE|100|RELU,DENSE|1|LINEAR" [Layout of the network.]
: ErrorStrategy: "CROSSENTROPY" [Loss function: Mean squared error (regression) or cross entropy (binary classification).]
: WeightInitialization: "XAVIER" [Weight initialization strategy]
: Architecture: "CPU" [Which architecture to perform the training on.]
: TrainingStrategy: "LearningRate=1e-3,Momentum=0.9,Repetitions=1,ConvergenceSteps=5,BatchSize=100,TestRepetitions=1,WeightDecay=1e-4,Regularization=None,Optimizer=ADAM,DropConfig=0.0+0.0+0.0+0.,MaxEpochs=10" [Defines the training strategies.]
: - Default:
: VerbosityLevel: "Default" [Verbosity level]
: CreateMVAPdfs: "False" [Create PDFs for classifier outputs (signal and background)]
: IgnoreNegWeightsInTraining: "False" [Events with negative weights are ignored in the training (but are included for testing and performance evaluation)]
: InputLayout: "0|0|0" [The Layout of the input]
: BatchLayout: "0|0|0" [The Layout of the batch]
: RandomSeed: "0" [Random seed used for weight initialization and batch shuffling]
: ValidationSize: "20%" [Part of the training data to use for validation. Specify as 0.2 or 20% to use a fifth of the data set as validation set. Specify as 100 to use exactly 100 events. (Default: 20%)]
: Will now use the CPU architecture with BLAS and IMT support !
Factory : Booking method: ␛[1mTMVA_CNN_CPU␛[0m
:
: Parsing option string:
: ... "!H:V:ErrorStrategy=CROSSENTROPY:VarTransform=None:WeightInitialization=XAVIER:InputLayout=1|16|16:Layout=CONV|10|3|3|1|1|1|1|RELU,BNORM,CONV|10|3|3|1|1|1|1|RELU,MAXPOOL|2|2|1|1,RESHAPE|FLAT,DENSE|100|RELU,DENSE|1|LINEAR:TrainingStrategy=LearningRate=1e-3,Momentum=0.9,Repetitions=1,ConvergenceSteps=5,BatchSize=100,TestRepetitions=1,WeightDecay=1e-4,Regularization=None,Optimizer=ADAM,DropConfig=0.0+0.0+0.0+0.0,MaxEpochs=10:Architecture=CPU"
: The following options are set:
: - By User:
: <none>
: - Default:
: Boost_num: "0" [Number of times the classifier will be boosted]
: Parsing option string:
: ... "!H:V:ErrorStrategy=CROSSENTROPY:VarTransform=None:WeightInitialization=XAVIER:InputLayout=1|16|16:Layout=CONV|10|3|3|1|1|1|1|RELU,BNORM,CONV|10|3|3|1|1|1|1|RELU,MAXPOOL|2|2|1|1,RESHAPE|FLAT,DENSE|100|RELU,DENSE|1|LINEAR:TrainingStrategy=LearningRate=1e-3,Momentum=0.9,Repetitions=1,ConvergenceSteps=5,BatchSize=100,TestRepetitions=1,WeightDecay=1e-4,Regularization=None,Optimizer=ADAM,DropConfig=0.0+0.0+0.0+0.0,MaxEpochs=10:Architecture=CPU"
: The following options are set:
: - By User:
: V: "True" [Verbose output (short form of "VerbosityLevel" below - overrides the latter one)]
: VarTransform: "None" [List of variable transformations performed before training, e.g., "D_Background,P_Signal,G,N_AllClasses" for: "Decorrelation, PCA-transformation, Gaussianisation, Normalisation, each for the given class of events ('AllClasses' denotes all events of all classes, if no class indication is given, 'All' is assumed)"]
: H: "False" [Print method-specific help message]
: InputLayout: "1|16|16" [The Layout of the input]
: Layout: "CONV|10|3|3|1|1|1|1|RELU,BNORM,CONV|10|3|3|1|1|1|1|RELU,MAXPOOL|2|2|1|1,RESHAPE|FLAT,DENSE|100|RELU,DENSE|1|LINEAR" [Layout of the network.]
: ErrorStrategy: "CROSSENTROPY" [Loss function: Mean squared error (regression) or cross entropy (binary classification).]
: WeightInitialization: "XAVIER" [Weight initialization strategy]
: Architecture: "CPU" [Which architecture to perform the training on.]
: TrainingStrategy: "LearningRate=1e-3,Momentum=0.9,Repetitions=1,ConvergenceSteps=5,BatchSize=100,TestRepetitions=1,WeightDecay=1e-4,Regularization=None,Optimizer=ADAM,DropConfig=0.0+0.0+0.0+0.0,MaxEpochs=10" [Defines the training strategies.]
: - Default:
: VerbosityLevel: "Default" [Verbosity level]
: CreateMVAPdfs: "False" [Create PDFs for classifier outputs (signal and background)]
: IgnoreNegWeightsInTraining: "False" [Events with negative weights are ignored in the training (but are included for testing and performance evaluation)]
: BatchLayout: "0|0|0" [The Layout of the batch]
: RandomSeed: "0" [Random seed used for weight initialization and batch shuffling]
: ValidationSize: "20%" [Part of the training data to use for validation. Specify as 0.2 or 20% to use a fifth of the data set as validation set. Specify as 100 to use exactly 100 events. (Default: 20%)]
: Will now use the CPU architecture with BLAS and IMT support !
Factory : ␛[1mTrain all methods␛[0m
Factory : Train method: BDT for Classification
:
BDT : #events: (reweighted) sig: 800 bkg: 800
: #events: (unweighted) sig: 800 bkg: 800
: Training 400 Decision Trees ... patience please
: Elapsed time for training with 1600 events: 1.71 sec
BDT : [dataset] : Evaluation of BDT on training sample (1600 events)
: Elapsed time for evaluation of 1600 events: 0.0164 sec
: Creating xml weight file: ␛[0;36mdataset/weights/TMVA_CNN_Classification_BDT.weights.xml␛[0m
: Creating standalone class: ␛[0;36mdataset/weights/TMVA_CNN_Classification_BDT.class.C␛[0m
: TMVA_CNN_ClassificationOutput.root:/dataset/Method_BDT/BDT
Factory : Training finished
:
Factory : Train method: TMVA_DNN_CPU for Classification
:
: Start of deep neural network training on CPU using MT, nthreads = 4
:
: ***** Deep Learning Network *****
DEEP NEURAL NETWORK: Depth = 8 Input = ( 1, 1, 256 ) Batch size = 100 Loss function = C
Layer 0 DENSE Layer: ( Input = 256 , Width = 100 ) Output = ( 1 , 100 , 100 ) Activation Function = Relu
Layer 1 BATCH NORM Layer: Input/Output = ( 100 , 100 , 1 ) Norm dim = 100 axis = -1
Layer 2 DENSE Layer: ( Input = 100 , Width = 100 ) Output = ( 1 , 100 , 100 ) Activation Function = Relu
Layer 3 BATCH NORM Layer: Input/Output = ( 100 , 100 , 1 ) Norm dim = 100 axis = -1
Layer 4 DENSE Layer: ( Input = 100 , Width = 100 ) Output = ( 1 , 100 , 100 ) Activation Function = Relu
Layer 5 BATCH NORM Layer: Input/Output = ( 100 , 100 , 1 ) Norm dim = 100 axis = -1
Layer 6 DENSE Layer: ( Input = 100 , Width = 100 ) Output = ( 1 , 100 , 100 ) Activation Function = Relu
Layer 7 DENSE Layer: ( Input = 100 , Width = 1 ) Output = ( 1 , 100 , 1 ) Activation Function = Identity
: Using 1280 events for training and 320 for testing
: Compute initial loss on the validation data
: Training phase 1 of 1: Optimizer ADAM (beta1=0.9,beta2=0.999,eps=1e-07) Learning rate = 0.001 regularization 0 minimum error = 21.8273
: --------------------------------------------------------------
: Epoch | Train Err. Val. Err. t(s)/epoch t(s)/Loss nEvents/s Conv. Steps
: --------------------------------------------------------------
: Start epoch iteration ...
: 1 Minimum Test error found - save the configuration
: 1 | 1.0348 1.50777 0.147606 0.0187439 9312.28 0
: 2 Minimum Test error found - save the configuration
: 2 | 0.700302 1.27075 0.176006 0.019543 7669.53 0
: 3 Minimum Test error found - save the configuration
: 3 | 0.625502 1.12068 0.169952 0.0188981 7944.18 0
: 4 Minimum Test error found - save the configuration
: 4 | 0.561372 1.11073 0.176747 0.0200042 7655.85 0
: 5 Minimum Test error found - save the configuration
: 5 | 0.496457 1.09581 0.174257 0.0192665 7742.44 0
: 6 | 0.431827 1.14041 0.172394 0.0184929 7797.22 1
: 7 | 0.397026 1.15564 0.171003 0.0189866 7893.87 2
: 8 Minimum Test error found - save the configuration
: 8 | 0.342025 1.09064 0.170182 0.019092 7942.26 0
: 9 | 0.286433 1.18584 0.171795 0.018873 7847.15 1
: 10 | 0.261693 1.2418 0.164411 0.0174628 8166.14 2
:
: Elapsed time for training with 1600 events: 1.72 sec
: Evaluate deep neural network on CPU using batches with size = 100
:
TMVA_DNN_CPU : [dataset] : Evaluation of TMVA_DNN_CPU on training sample (1600 events)
: Elapsed time for evaluation of 1600 events: 0.102 sec
: Creating xml weight file: ␛[0;36mdataset/weights/TMVA_CNN_Classification_TMVA_DNN_CPU.weights.xml␛[0m
: Creating standalone class: ␛[0;36mdataset/weights/TMVA_CNN_Classification_TMVA_DNN_CPU.class.C␛[0m
Factory : Training finished
:
Factory : Train method: TMVA_CNN_CPU for Classification
:
: Start of deep neural network training on CPU using MT, nthreads = 4
:
: ***** Deep Learning Network *****
DEEP NEURAL NETWORK: Depth = 7 Input = ( 1, 16, 16 ) Batch size = 100 Loss function = C
Layer 0 CONV LAYER: ( W = 16 , H = 16 , D = 10 ) Filter ( W = 3 , H = 3 ) Output = ( 100 , 10 , 10 , 256 ) Activation Function = Relu
Layer 1 BATCH NORM Layer: Input/Output = ( 10 , 256 , 100 ) Norm dim = 10 axis = 1
Layer 2 CONV LAYER: ( W = 16 , H = 16 , D = 10 ) Filter ( W = 3 , H = 3 ) Output = ( 100 , 10 , 10 , 256 ) Activation Function = Relu
Layer 3 POOL Layer: ( W = 15 , H = 15 , D = 10 ) Filter ( W = 2 , H = 2 ) Output = ( 100 , 10 , 10 , 225 )
Layer 4 RESHAPE Layer Input = ( 10 , 15 , 15 ) Output = ( 1 , 100 , 2250 )
Layer 5 DENSE Layer: ( Input = 2250 , Width = 100 ) Output = ( 1 , 100 , 100 ) Activation Function = Relu
Layer 6 DENSE Layer: ( Input = 100 , Width = 1 ) Output = ( 1 , 100 , 1 ) Activation Function = Identity
: Using 1280 events for training and 320 for testing
: Compute initial loss on the validation data
: Training phase 1 of 1: Optimizer ADAM (beta1=0.9,beta2=0.999,eps=1e-07) Learning rate = 0.001 regularization 0 minimum error = 184.928
: --------------------------------------------------------------
: Epoch | Train Err. Val. Err. t(s)/epoch t(s)/Loss nEvents/s Conv. Steps
: --------------------------------------------------------------
: Start epoch iteration ...
: 1 Minimum Test error found - save the configuration
: 1 | 6.2677 3.35653 1.01181 0.0837096 1292.96 0
: 2 Minimum Test error found - save the configuration
: 2 | 1.70727 0.732628 0.918189 0.0875828 1444.73 0
: 3 | 0.890542 0.750677 0.942725 0.0853785 1399.67 1
: 4 Minimum Test error found - save the configuration
: 4 | 0.704229 0.686011 1.15176 0.117186 1159.9 0
: 5 Minimum Test error found - save the configuration
: 5 | 0.671897 0.67311 0.923088 0.081224 1425.41 0
: 6 | 0.664014 0.710039 1.02978 0.111701 1307.07 1
: 7 Minimum Test error found - save the configuration
: 7 | 0.652514 0.636182 1.04979 0.0881565 1247.87 0
: 8 Minimum Test error found - save the configuration
: 8 | 0.634474 0.626774 0.944327 0.0841509 1395.06 0
: 9 Minimum Test error found - save the configuration
: 9 | 0.66644 0.611112 1.02156 0.0837529 1279.58 0
: 10 Minimum Test error found - save the configuration
: 10 | 0.589074 0.595838 1.03715 0.114535 1300.66 0
:
: Elapsed time for training with 1600 events: 10.2 sec
: Evaluate deep neural network on CPU using batches with size = 100
:
TMVA_CNN_CPU : [dataset] : Evaluation of TMVA_CNN_CPU on training sample (1600 events)
: Elapsed time for evaluation of 1600 events: 0.525 sec
: Creating xml weight file: ␛[0;36mdataset/weights/TMVA_CNN_Classification_TMVA_CNN_CPU.weights.xml␛[0m
: Creating standalone class: ␛[0;36mdataset/weights/TMVA_CNN_Classification_TMVA_CNN_CPU.class.C␛[0m
Factory : Training finished
:
: Ranking input variables (method specific)...
BDT : Ranking result (top variable is best ranked)
: --------------------------------------
: Rank : Variable : Variable Importance
: --------------------------------------
: 1 : vars : 8.150e-03
: 2 : vars : 7.852e-03
: 3 : vars : 7.321e-03
: 4 : vars : 7.018e-03
: 5 : vars : 6.905e-03
: 6 : vars : 6.821e-03
: 7 : vars : 6.733e-03
: 8 : vars : 6.710e-03
: 9 : vars : 6.660e-03
: 10 : vars : 6.587e-03
: 11 : vars : 6.575e-03
: 12 : vars : 6.547e-03
: 13 : vars : 6.516e-03
: 14 : vars : 6.492e-03
: 15 : vars : 6.434e-03
: 16 : vars : 6.382e-03
: 17 : vars : 6.381e-03
: 18 : vars : 6.377e-03
: 19 : vars : 6.357e-03
: 20 : vars : 6.160e-03
: 21 : vars : 6.146e-03
: 22 : vars : 6.119e-03
: 23 : vars : 6.116e-03
: 24 : vars : 6.062e-03
: 25 : vars : 6.059e-03
: 26 : vars : 5.951e-03
: 27 : vars : 5.893e-03
: 28 : vars : 5.858e-03
: 29 : vars : 5.815e-03
: 30 : vars : 5.815e-03
: 31 : vars : 5.809e-03
: 32 : vars : 5.804e-03
: 33 : vars : 5.800e-03
: 34 : vars : 5.788e-03
: 35 : vars : 5.768e-03
: 36 : vars : 5.764e-03
: 37 : vars : 5.709e-03
: 38 : vars : 5.663e-03
: 39 : vars : 5.642e-03
: 40 : vars : 5.596e-03
: 41 : vars : 5.596e-03
: 42 : vars : 5.578e-03
: 43 : vars : 5.523e-03
: 44 : vars : 5.474e-03
: 45 : vars : 5.462e-03
: 46 : vars : 5.413e-03
: 47 : vars : 5.381e-03
: 48 : vars : 5.363e-03
: 49 : vars : 5.337e-03
: 50 : vars : 5.316e-03
: 51 : vars : 5.279e-03
: 52 : vars : 5.274e-03
: 53 : vars : 5.269e-03
: 54 : vars : 5.246e-03
: 55 : vars : 5.242e-03
: 56 : vars : 5.212e-03
: 57 : vars : 5.175e-03
: 58 : vars : 5.173e-03
: 59 : vars : 5.139e-03
: 60 : vars : 5.133e-03
: 61 : vars : 5.091e-03
: 62 : vars : 5.068e-03
: 63 : vars : 5.050e-03
: 64 : vars : 5.037e-03
: 65 : vars : 5.000e-03
: 66 : vars : 4.993e-03
: 67 : vars : 4.977e-03
: 68 : vars : 4.941e-03
: 69 : vars : 4.913e-03
: 70 : vars : 4.906e-03
: 71 : vars : 4.898e-03
: 72 : vars : 4.838e-03
: 73 : vars : 4.831e-03
: 74 : vars : 4.829e-03
: 75 : vars : 4.817e-03
: 76 : vars : 4.779e-03
: 77 : vars : 4.771e-03
: 78 : vars : 4.745e-03
: 79 : vars : 4.709e-03
: 80 : vars : 4.670e-03
: 81 : vars : 4.666e-03
: 82 : vars : 4.646e-03
: 83 : vars : 4.624e-03
: 84 : vars : 4.620e-03
: 85 : vars : 4.573e-03
: 86 : vars : 4.572e-03
: 87 : vars : 4.549e-03
: 88 : vars : 4.549e-03
: 89 : vars : 4.523e-03
: 90 : vars : 4.513e-03
: 91 : vars : 4.510e-03
: 92 : vars : 4.501e-03
: 93 : vars : 4.491e-03
: 94 : vars : 4.475e-03
: 95 : vars : 4.456e-03
: 96 : vars : 4.453e-03
: 97 : vars : 4.438e-03
: 98 : vars : 4.429e-03
: 99 : vars : 4.399e-03
: 100 : vars : 4.343e-03
: 101 : vars : 4.328e-03
: 102 : vars : 4.300e-03
: 103 : vars : 4.286e-03
: 104 : vars : 4.251e-03
: 105 : vars : 4.242e-03
: 106 : vars : 4.239e-03
: 107 : vars : 4.231e-03
: 108 : vars : 4.224e-03
: 109 : vars : 4.212e-03
: 110 : vars : 4.197e-03
: 111 : vars : 4.188e-03
: 112 : vars : 4.161e-03
: 113 : vars : 4.116e-03
: 114 : vars : 4.109e-03
: 115 : vars : 4.094e-03
: 116 : vars : 4.090e-03
: 117 : vars : 4.085e-03
: 118 : vars : 4.072e-03
: 119 : vars : 4.062e-03
: 120 : vars : 4.036e-03
: 121 : vars : 4.014e-03
: 122 : vars : 4.001e-03
: 123 : vars : 3.993e-03
: 124 : vars : 3.979e-03
: 125 : vars : 3.978e-03
: 126 : vars : 3.968e-03
: 127 : vars : 3.967e-03
: 128 : vars : 3.947e-03
: 129 : vars : 3.947e-03
: 130 : vars : 3.932e-03
: 131 : vars : 3.915e-03
: 132 : vars : 3.909e-03
: 133 : vars : 3.888e-03
: 134 : vars : 3.882e-03
: 135 : vars : 3.878e-03
: 136 : vars : 3.854e-03
: 137 : vars : 3.837e-03
: 138 : vars : 3.836e-03
: 139 : vars : 3.829e-03
: 140 : vars : 3.826e-03
: 141 : vars : 3.803e-03
: 142 : vars : 3.796e-03
: 143 : vars : 3.791e-03
: 144 : vars : 3.785e-03
: 145 : vars : 3.766e-03
: 146 : vars : 3.748e-03
: 147 : vars : 3.713e-03
: 148 : vars : 3.677e-03
: 149 : vars : 3.664e-03
: 150 : vars : 3.655e-03
: 151 : vars : 3.652e-03
: 152 : vars : 3.650e-03
: 153 : vars : 3.619e-03
: 154 : vars : 3.618e-03
: 155 : vars : 3.611e-03
: 156 : vars : 3.577e-03
: 157 : vars : 3.576e-03
: 158 : vars : 3.549e-03
: 159 : vars : 3.517e-03
: 160 : vars : 3.502e-03
: 161 : vars : 3.484e-03
: 162 : vars : 3.477e-03
: 163 : vars : 3.475e-03
: 164 : vars : 3.445e-03
: 165 : vars : 3.405e-03
: 166 : vars : 3.390e-03
: 167 : vars : 3.349e-03
: 168 : vars : 3.347e-03
: 169 : vars : 3.341e-03
: 170 : vars : 3.338e-03
: 171 : vars : 3.283e-03
: 172 : vars : 3.281e-03
: 173 : vars : 3.255e-03
: 174 : vars : 3.225e-03
: 175 : vars : 3.212e-03
: 176 : vars : 3.192e-03
: 177 : vars : 3.145e-03
: 178 : vars : 3.142e-03
: 179 : vars : 3.140e-03
: 180 : vars : 3.134e-03
: 181 : vars : 3.112e-03
: 182 : vars : 3.105e-03
: 183 : vars : 3.102e-03
: 184 : vars : 3.088e-03
: 185 : vars : 3.059e-03
: 186 : vars : 3.056e-03
: 187 : vars : 3.048e-03
: 188 : vars : 3.028e-03
: 189 : vars : 3.026e-03
: 190 : vars : 3.025e-03
: 191 : vars : 3.025e-03
: 192 : vars : 2.991e-03
: 193 : vars : 2.960e-03
: 194 : vars : 2.857e-03
: 195 : vars : 2.834e-03
: 196 : vars : 2.833e-03
: 197 : vars : 2.816e-03
: 198 : vars : 2.807e-03
: 199 : vars : 2.790e-03
: 200 : vars : 2.760e-03
: 201 : vars : 2.759e-03
: 202 : vars : 2.750e-03
: 203 : vars : 2.739e-03
: 204 : vars : 2.675e-03
: 205 : vars : 2.669e-03
: 206 : vars : 2.627e-03
: 207 : vars : 2.609e-03
: 208 : vars : 2.576e-03
: 209 : vars : 2.574e-03
: 210 : vars : 2.573e-03
: 211 : vars : 2.554e-03
: 212 : vars : 2.496e-03
: 213 : vars : 2.477e-03
: 214 : vars : 2.440e-03
: 215 : vars : 2.402e-03
: 216 : vars : 2.379e-03
: 217 : vars : 2.326e-03
: 218 : vars : 2.293e-03
: 219 : vars : 2.279e-03
: 220 : vars : 2.187e-03
: 221 : vars : 2.158e-03
: 222 : vars : 2.143e-03
: 223 : vars : 2.102e-03
: 224 : vars : 2.088e-03
: 225 : vars : 2.081e-03
: 226 : vars : 2.062e-03
: 227 : vars : 1.939e-03
: 228 : vars : 1.921e-03
: 229 : vars : 1.903e-03
: 230 : vars : 1.849e-03
: 231 : vars : 1.828e-03
: 232 : vars : 1.705e-03
: 233 : vars : 1.705e-03
: 234 : vars : 1.701e-03
: 235 : vars : 1.693e-03
: 236 : vars : 1.649e-03
: 237 : vars : 1.624e-03
: 238 : vars : 1.555e-03
: 239 : vars : 1.429e-03
: 240 : vars : 1.417e-03
: 241 : vars : 1.361e-03
: 242 : vars : 1.345e-03
: 243 : vars : 1.220e-03
: 244 : vars : 1.039e-03
: 245 : vars : 6.119e-04
: 246 : vars : 0.000e+00
: 247 : vars : 0.000e+00
: 248 : vars : 0.000e+00
: 249 : vars : 0.000e+00
: 250 : vars : 0.000e+00
: 251 : vars : 0.000e+00
: 252 : vars : 0.000e+00
: 253 : vars : 0.000e+00
: 254 : vars : 0.000e+00
: 255 : vars : 0.000e+00
: 256 : vars : 0.000e+00
: --------------------------------------
: No variable ranking supplied by classifier: TMVA_DNN_CPU
: No variable ranking supplied by classifier: TMVA_CNN_CPU
TH1.Print Name = TrainingHistory_TMVA_DNN_CPU_trainingError, Entries= 0, Total sum= 5.13744
TH1.Print Name = TrainingHistory_TMVA_DNN_CPU_valError, Entries= 0, Total sum= 11.9201
TH1.Print Name = TrainingHistory_TMVA_CNN_CPU_trainingError, Entries= 0, Total sum= 13.4482
TH1.Print Name = TrainingHistory_TMVA_CNN_CPU_valError, Entries= 0, Total sum= 9.37891
Factory : === Destroy and recreate all methods via weight files for testing ===
:
: Reading weight file: ␛[0;36mdataset/weights/TMVA_CNN_Classification_BDT.weights.xml␛[0m
: Reading weight file: ␛[0;36mdataset/weights/TMVA_CNN_Classification_TMVA_DNN_CPU.weights.xml␛[0m
: Reading weight file: ␛[0;36mdataset/weights/TMVA_CNN_Classification_TMVA_CNN_CPU.weights.xml␛[0m
Factory : ␛[1mTest all methods␛[0m
Factory : Test method: BDT for Classification performance
:
BDT : [dataset] : Evaluation of BDT on testing sample (400 events)
: Elapsed time for evaluation of 400 events: 0.00466 sec
Factory : Test method: TMVA_DNN_CPU for Classification performance
:
: Evaluate deep neural network on CPU using batches with size = 400
:
TMVA_DNN_CPU : [dataset] : Evaluation of TMVA_DNN_CPU on testing sample (400 events)
: Elapsed time for evaluation of 400 events: 0.0251 sec
Factory : Test method: TMVA_CNN_CPU for Classification performance
:
: Evaluate deep neural network on CPU using batches with size = 400
:
TMVA_CNN_CPU : [dataset] : Evaluation of TMVA_CNN_CPU on testing sample (400 events)
: Elapsed time for evaluation of 400 events: 0.128 sec
Factory : ␛[1mEvaluate all methods␛[0m
Factory : Evaluate classifier: BDT
:
BDT : [dataset] : Loop over test events and fill histograms with classifier response...
:
: Dataset[dataset] : variable plots are not produces ! The number of variables is 256 , it is larger than 200
Factory : Evaluate classifier: TMVA_DNN_CPU
:
TMVA_DNN_CPU : [dataset] : Loop over test events and fill histograms with classifier response...
:
: Evaluate deep neural network on CPU using batches with size = 1000
:
: Dataset[dataset] : variable plots are not produces ! The number of variables is 256 , it is larger than 200
Factory : Evaluate classifier: TMVA_CNN_CPU
:
TMVA_CNN_CPU : [dataset] : Loop over test events and fill histograms with classifier response...
:
: Evaluate deep neural network on CPU using batches with size = 1000
:
: Dataset[dataset] : variable plots are not produces ! The number of variables is 256 , it is larger than 200
:
: Evaluation results ranked by best signal efficiency and purity (area)
: -------------------------------------------------------------------------------------------------------------------
: DataSet MVA
: Name: Method: ROC-integ
: dataset TMVA_CNN_CPU : 0.756
: dataset BDT : 0.744
: dataset TMVA_DNN_CPU : 0.692
: -------------------------------------------------------------------------------------------------------------------
:
: Testing efficiency compared to training efficiency (overtraining check)
: -------------------------------------------------------------------------------------------------------------------
: DataSet MVA Signal efficiency: from test sample (from training sample)
: Name: Method: @B=0.01 @B=0.10 @B=0.30
: -------------------------------------------------------------------------------------------------------------------
: dataset TMVA_CNN_CPU : 0.025 (0.095) 0.365 (0.411) 0.680 (0.748)
: dataset BDT : 0.045 (0.375) 0.345 (0.709) 0.638 (0.883)
: dataset TMVA_DNN_CPU : 0.025 (0.105) 0.305 (0.545) 0.600 (0.736)
: -------------------------------------------------------------------------------------------------------------------
:
Dataset:dataset : Created tree 'TestTree' with 400 events
:
Dataset:dataset : Created tree 'TrainTree' with 1600 events
:
Factory : ␛[1mThank you for using TMVA!␛[0m
: ␛[1mFor citation information, please visit: http://tmva.sf.net/citeTMVA.html␛[0m
# TMVA Classification Example Using a Convolutional Neural Network
## Helper function to create input images data
## we create a signal and background 2D histograms from 2d gaussians
## with a location (means in X and Y) different for each event
## The difference between signal and background is in the gaussian width.
## The width for the background gaussian is slightly larger than the signal width by few % values
import ROOT
import os
opt = [1, 1, 1, 1, 1]
useTMVACNN = opt[0] if len(opt) > 0 else False
useKerasCNN = opt[1] if len(opt) > 1 else False
useTMVADNN = opt[2] if len(opt) > 2 else False
useTMVABDT = opt[3] if len(opt) > 3 else False
usePyTorchCNN = opt[4] if len(opt) > 4 else False
TMVA = ROOT.TMVA
TFile = ROOT.TFile
def MakeImagesTree(n, nh, nw):
# image size (nh x nw)
ntot = nh * nw
fileOutName = "images_data_16x16.root"
nRndmEvts = 10000 # number of events we use to fill each image
delta_sigma = 0.1 # 5% difference in the sigma
pixelNoise = 5
sX1 = 3
sY1 = 3
sX2 = sX1 + delta_sigma
sY2 = sY1 - delta_sigma
h1 = ROOT.TH2D("h1", "h1", nh, 0, 10, nw, 0, 10)
h2 = ROOT.TH2D("h2", "h2", nh, 0, 10, nw, 0, 10)
f1 = ROOT.TF2("f1", "xygaus")
f2 = ROOT.TF2("f2", "xygaus")
sgn = ROOT.TTree("sig_tree", "signal_tree")
bkg = ROOT.TTree("bkg_tree", "background_tree")
f = TFile(fileOutName, "RECREATE")
x1 = ROOT.std.vector["float"](ntot)
x2 = ROOT.std.vector["float"](ntot)
# create signal and background trees with a single branch
# an std::vector<float> of size nh x nw containing the image data
bkg.Branch("vars", "std::vector<float>", x1)
sgn.Branch("vars", "std::vector<float>", x2)
f1.SetParameters(1, 5, sX1, 5, sY1)
f2.SetParameters(1, 5, sX2, 5, sY2)
ROOT.Info("TMVA_CNN_Classification", "Filling ROOT tree \n")
for i in range(n):
if i % 1000 == 0:
print("Generating image event ...", i)
# generate random means in range [3,7] to be not too much on the border
h1.FillRandom(f1, nRndmEvts)
h2.FillRandom(f2, nRndmEvts)
for k in range(nh):
for l in range(nw):
m = k * nw + l
# add some noise in each bin
x1[m] = h1.GetBinContent(k + 1, l + 1) + ROOT.gRandom.Gaus(0, pixelNoise)
x2[m] = h2.GetBinContent(k + 1, l + 1) + ROOT.gRandom.Gaus(0, pixelNoise)
print("Signal and background tree with images data written to the file %s", f.GetName())
hasGPU = "tmva-gpu" in ROOT.gROOT.GetConfigFeatures()
hasCPU = "tmva-cpu" in ROOT.gROOT.GetConfigFeatures()
nevt = 1000 # use a larger value to get better results
if (not hasCPU and not hasGPU) :
ROOT.Warning("TMVA_CNN_Classificaton","ROOT is not supporting tmva-cpu and tmva-gpu skip using TMVA-DNN and TMVA-CNN")
useTMVACNN = False
useTMVADNN = False
if not "tmva-pymva" in ROOT.gROOT.GetConfigFeatures():
useKerasCNN = False
usePyTorchCNN = False
else:
if not useTMVACNN:
"TMVA_CNN_Classificaton",
"TMVA is not build with GPU or CPU multi-thread support. Cannot use TMVA Deep Learning for CNN",
)
writeOutputFile = True
num_threads = 4 # use max 4 threads
max_epochs = 10 # maximum number of epochs used for training
# do enable MT running
ROOT.EnableImplicitMT(num_threads)
ROOT.gSystem.Setenv("OMP_NUM_THREADS", "1") # switch OFF MT in OpenBLAS
print("Running with nthreads = {}".format(ROOT.GetThreadPoolSize()))
else:
print("Running in serial mode since ROOT does not support MT")
outputFile = None
if writeOutputFile:
outputFile = TFile.Open("TMVA_CNN_ClassificationOutput.root", "RECREATE")
## Create TMVA Factory
# Create the Factory class. Later you can choose the methods
# whose performance you'd like to investigate.
# The factory is the major TMVA object you have to interact with. Here is the list of parameters you need to pass
# - The first argument is the base of the name of all the output
# weight files in the directory weight/ that will be created with the
# method parameters
# - The second argument is the output file for the training results
# - The third argument is a string option defining some general configuration for the TMVA session.
# For example all TMVA output can be suppressed by removing the "!" (not) in front of the "Silent" argument in the
# option string
# - note that we disable any pre-transformation of the input variables and we avoid computing correlations between
# input variables
factory = TMVA.Factory(
"TMVA_CNN_Classification",
outputFile,
V=False,
ROC=True,
Silent=False,
Color=True,
AnalysisType="Classification",
Transformations=None,
Correlations=False,
)
## Declare DataLoader(s)
# The next step is to declare the DataLoader class that deals with input variables
# Define the input variables that shall be used for the MVA training
# note that you may also use variable expressions, which can be parsed by TTree::Draw( "expression" )]
# In this case the input data consists of an image of 16x16 pixels. Each single pixel is a branch in a ROOT TTree
loader = TMVA.DataLoader("dataset")
## Setup Dataset(s)
# Define input data file and signal and background trees
imgSize = 16 * 16
inputFileName = "images_data_16x16.root"
# if the input file does not exist create it
if ROOT.gSystem.AccessPathName(inputFileName):
MakeImagesTree(nevt, 16, 16)
inputFile = TFile.Open(inputFileName)
if inputFile is None:
ROOT.Warning("TMVA_CNN_Classification", "Error opening input file %s - exit", inputFileName.Data())
# inputFileName = "tmva_class_example.root"
# --- Register the training and test trees
signalTree = inputFile.Get("sig_tree")
backgroundTree = inputFile.Get("bkg_tree")
nEventsSig = signalTree.GetEntries()
# global event weights per tree (see below for setting event-wise weights)
signalWeight = 1.0
backgroundWeight = 1.0
# You can add an arbitrary number of signal or background trees
loader.AddSignalTree(signalTree, signalWeight)
loader.AddBackgroundTree(backgroundTree, backgroundWeight)
## add event variables (image)
## use new method (from ROOT 6.20 to add a variable array for all image data)
loader.AddVariablesArray("vars", imgSize)
# Set individual event weights (the variables must exist in the original TTree)
# for signal : factory->SetSignalWeightExpression ("weight1*weight2");
# for background: factory->SetBackgroundWeightExpression("weight1*weight2");
# loader->SetBackgroundWeightExpression( "weight" );
# Apply additional cuts on the signal and background samples (can be different)
mycuts = "" # for example: TCut mycuts = "abs(var1)<0.5 && abs(var2-0.5)<1";
mycutb = "" # for example: TCut mycutb = "abs(var1)<0.5";
# Tell the factory how to use the training and testing events
# If no numbers of events are given, half of the events in the tree are used
# for training, and the other half for testing:
# loader.PrepareTrainingAndTestTree( mycut, "SplitMode=random:!V" );
# It is possible also to specify the number of training and testing events,
# note we disable the computation of the correlation matrix of the input variables
nTrainSig = 0.8 * nEventsSig
nTrainBkg = 0.8 * nEventsBkg
# build the string options for DataLoader::PrepareTrainingAndTestTree
mycuts,
mycutb,
nTrain_Signal=nTrainSig,
nTrain_Background=nTrainBkg,
SplitMode="Random",
SplitSeed=100,
NormMode="NumEvents",
V=False,
CalcCorrelations=False,
)
# DataSetInfo : [dataset] : Added class "Signal"
# : Add Tree sig_tree of type Signal with 10000 events
# DataSetInfo : [dataset] : Added class "Background"
# : Add Tree bkg_tree of type Background with 10000 events
# signalTree.Print();
# Booking Methods
# Here we book the TMVA methods. We book a Boosted Decision Tree method (BDT)
# Boosted Decision Trees
if useTMVABDT:
loader,
"BDT",
V=False,
NTrees=400,
MinNodeSize="2.5%",
MaxDepth=2,
BoostType="AdaBoost",
AdaBoostBeta=0.5,
UseBaggedBoost=True,
BaggedSampleFraction=0.5,
SeparationType="GiniIndex",
nCuts=20,
)
#### Booking Deep Neural Network
# Here we book the DNN of TMVA. See the example TMVA_Higgs_Classification.C for a detailed description of the
# options
if useTMVADNN:
layoutString = ROOT.TString(
"DENSE|100|RELU,BNORM,DENSE|100|RELU,BNORM,DENSE|100|RELU,BNORM,DENSE|100|RELU,DENSE|1|LINEAR"
)
# Training strategies
# one can catenate several training strings with different parameters (e.g. learning rates or regularizations
# parameters) The training string must be concatenated with the `|` delimiter
trainingString1 = ROOT.TString(
"LearningRate=1e-3,Momentum=0.9,Repetitions=1,"
"ConvergenceSteps=5,BatchSize=100,TestRepetitions=1,"
"WeightDecay=1e-4,Regularization=None,"
"Optimizer=ADAM,DropConfig=0.0+0.0+0.0+0."
) # + "|" + trainingString2 + ...
trainingString1 += ",MaxEpochs=" + str(max_epochs)
# Build now the full DNN Option string
dnnMethodName = "TMVA_DNN_CPU"
# use GPU if available
dnnOptions = "CPU"
if hasGPU :
dnnOptions = "GPU"
dnnMethodName = "TMVA_DNN_GPU"
loader,
dnnMethodName,
H=False,
V=True,
ErrorStrategy="CROSSENTROPY",
VarTransform=None,
WeightInitialization="XAVIER",
Layout=layoutString,
TrainingStrategy=trainingString1,
Architecture=dnnOptions
)
### Book Convolutional Neural Network in TMVA
# For building a CNN one needs to define
# - Input Layout : number of channels (in this case = 1) | image height | image width
# - Batch Layout : batch size | number of channels | image size = (height*width)
# Then one add Convolutional layers and MaxPool layers.
# - For Convolutional layer the option string has to be:
# - CONV | number of units | filter height | filter width | stride height | stride width | padding height | paddig
# width | activation function
# - note in this case we are using a filer 3x3 and padding=1 and stride=1 so we get the output dimension of the
# conv layer equal to the input
# - note we use after the first convolutional layer a batch normalization layer. This seems to help significantly the
# convergence
# - For the MaxPool layer:
# - MAXPOOL | pool height | pool width | stride height | stride width
# The RESHAPE layer is needed to flatten the output before the Dense layer
# Note that to run the CNN is required to have CPU or GPU support
if useTMVACNN:
# Training strategies.
trainingString1 = ROOT.TString(
"LearningRate=1e-3,Momentum=0.9,Repetitions=1,"
"ConvergenceSteps=5,BatchSize=100,TestRepetitions=1,"
"WeightDecay=1e-4,Regularization=None,"
"Optimizer=ADAM,DropConfig=0.0+0.0+0.0+0.0"
)
trainingString1 += ",MaxEpochs=" + str(max_epochs)
## New DL (CNN)
cnnMethodName = "TMVA_CNN_CPU"
cnnOptions = "CPU"
# use GPU if available
if hasGPU:
cnnOptions = "GPU"
cnnMethodName = "TMVA_CNN_GPU"
loader,
cnnMethodName,
H=False,
V=True,
ErrorStrategy="CROSSENTROPY",
VarTransform=None,
WeightInitialization="XAVIER",
InputLayout="1|16|16",
Layout="CONV|10|3|3|1|1|1|1|RELU,BNORM,CONV|10|3|3|1|1|1|1|RELU,MAXPOOL|2|2|1|1,RESHAPE|FLAT,DENSE|100|RELU,DENSE|1|LINEAR",
TrainingStrategy=trainingString1,
Architecture=cnnOptions,
)
### Book Convolutional Neural Network in Keras using a generated model
if usePyTorchCNN:
ROOT.Info("TMVA_CNN_Classification", "Using Convolutional PyTorch Model")
pyTorchFileName = str(ROOT.gROOT.GetTutorialDir())
pyTorchFileName += "/machine_learning/PyTorch_Generate_CNN_Model.py"
# check that pytorch can be imported and file defining the model exists
torch_spec = importlib.util.find_spec("torch")
if torch_spec is not None and os.path.exists(pyTorchFileName):
#cmd = str(ROOT.TMVA.Python_Executable()) + " " + pyTorchFileName
#os.system(cmd)
#import PyTorch_Generate_CNN_Model
ROOT.Info("TMVA_CNN_Classification", "Booking PyTorch CNN model")
loader,
"PyTorch",
H=True,
V=False,
VarTransform=None,
FilenameModel="PyTorchModelCNN.pt",
FilenameTrainedModel="PyTorchTrainedModelCNN.pt",
NumEpochs=max_epochs,
BatchSize=100,
UserCode=str(pyTorchFileName)
)
else:
"TMVA_CNN_Classification",
"PyTorch is not installed or model building file is not existing - skip using PyTorch",
)
if useKerasCNN:
ROOT.Info("TMVA_CNN_Classification", "Building convolutional keras model")
# create python script which can be executed
# create 2 conv2d layer + maxpool + dense
import tensorflow
from tensorflow.keras.models import Sequential
from tensorflow.keras.optimizers import Adam
# from keras.initializers import TruncatedNormal
# from keras import initializations
from tensorflow.keras.layers import Input, Dense, Dropout, Flatten, Conv2D, MaxPooling2D, Reshape
# from keras.callbacks import ReduceLROnPlateau
model = Sequential()
model.add(Reshape((16, 16, 1), input_shape=(256,)))
model.add(Conv2D(10, kernel_size=(3, 3), kernel_initializer="TruncatedNormal", activation="relu", padding="same"))
model.add(Conv2D(10, kernel_size=(3, 3), kernel_initializer="TruncatedNormal", activation="relu", padding="same"))
# stride for maxpool is equal to pool size
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Flatten())
model.add(Dense(64, activation="tanh"))
# model.add(Dropout(0.2))
model.add(Dense(2, activation="sigmoid"))
model.compile(loss="binary_crossentropy", optimizer=Adam(learning_rate=0.001), weighted_metrics=["accuracy"])
model.save("model_cnn.h5")
if not os.path.exists("model_cnn.h5"):
raise FileNotFoundError("Error creating Keras model file - skip using Keras")
else:
# book PyKeras method only if Keras model could be created
ROOT.Info("TMVA_CNN_Classification", "Booking convolutional keras model")
loader,
"PyKeras",
H=True,
V=False,
VarTransform=None,
FilenameModel="model_cnn.h5",
FilenameTrainedModel="trained_model_cnn.h5",
NumEpochs=max_epochs,
BatchSize=100,
GpuOptions="allow_growth=True",
) # needed for RTX NVidia card and to avoid TF allocates all GPU memory
## Train Methods
## Test and Evaluate Methods
## Plot ROC Curve
c1 = factory.GetROCCurve(loader)
# close outputfile to save output file
ROOT::Detail::TRangeCast< T, true > TRangeDynCast
TRangeDynCast is an adapter class that allows the typed iteration through a TCollection.
Option_t Option_t TPoint TPoint const char GetTextMagnitude GetFillStyle GetLineColor GetLineWidth GetMarkerStyle GetTextAlign GetTextColor GetTextSize void char Point_t Rectangle_t WindowAttributes_t Float_t Float_t Float_t Int_t Int_t UInt_t UInt_t Rectangle_t Int_t Int_t Window_t TString Int_t GCValues_t GetPrimarySelectionOwner GetDisplay GetScreen GetColormap GetNativeEvent const char const char dpyName wid window const char font_name cursor keysym reg const char only_if_exist regb h Point_t winding char text const char depth char const char Int_t count const char ColorStruct_t color const char Pixmap_t Pixmap_t PictureAttributes_t attr const char char ret_data h unsigned char height h Atom_t Int_t ULong_t ULong_t unsigned char prop_list Atom_t Atom_t Atom_t Time_t UChar_t len
Option_t Option_t TPoint TPoint const char GetTextMagnitude GetFillStyle GetLineColor GetLineWidth GetMarkerStyle GetTextAlign GetTextColor GetTextSize void char Point_t Rectangle_t WindowAttributes_t Float_t Float_t Float_t Int_t Int_t UInt_t UInt_t Rectangle_t Int_t Int_t Window_t TString Int_t GCValues_t GetPrimarySelectionOwner GetDisplay GetScreen GetColormap GetNativeEvent const char const char dpyName wid window const char font_name cursor keysym reg const char only_if_exist regb h Point_t winding char text const char depth char const char Int_t count const char ColorStruct_t color const char Pixmap_t Pixmap_t PictureAttributes_t attr const char char ret_data h unsigned char height h Atom_t Int_t ULong_t ULong_t unsigned char prop_list Atom_t Atom_t Atom_t Time_t format
A ROOT file is an on-disk file, usually with extension .root, that stores objects in a file-system-li...
Definition TFile.h:131
This is the main MVA steering class.
Definition Factory.h:80
Author
Harshal Shende

Definition in file TMVA_CNN_Classification.py.