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 = 16
DataSetInfo : [dataset] : Added class "Signal"
: Add Tree sig_tree of type Signal with 5000 events
DataSetInfo : [dataset] : Added class "Background"
: Add Tree bkg_tree of type Background with 5000 events
Factory : Booking method: ␛[1mBDT␛[0m
:
: 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 : 4000
: Signal -- testing events : 1000
: Signal -- training and testing events: 5000
: Background -- training events : 4000
: Background -- testing events : 1000
: Background -- training and testing events: 5000
:
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,MaxEpochs=20,WeightDecay=1e-4,Regularization=None,Optimizer=ADAM,DropConfig=0.0+0.0+0.0+0.: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,MaxEpochs=20,WeightDecay=1e-4,Regularization=None,Optimizer=ADAM,DropConfig=0.0+0.0+0.0+0.: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,MaxEpochs=20,WeightDecay=1e-4,Regularization=None,Optimizer=ADAM,DropConfig=0.0+0.0+0.0+0." [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,MaxEpochs=20,WeightDecay=1e-4,Regularization=None,Optimizer=ADAM,DropConfig=0.0+0.0+0.0+0.0: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,MaxEpochs=20,WeightDecay=1e-4,Regularization=None,Optimizer=ADAM,DropConfig=0.0+0.0+0.0+0.0: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,MaxEpochs=20,WeightDecay=1e-4,Regularization=None,Optimizer=ADAM,DropConfig=0.0+0.0+0.0+0.0" [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: 4000 bkg: 4000
: #events: (unweighted) sig: 4000 bkg: 4000
: Training 400 Decision Trees ... patience please
: Elapsed time for training with 8000 events: 7.89 sec
BDT : [dataset] : Evaluation of BDT on training sample (8000 events)
: Elapsed time for evaluation of 8000 events: 0.19 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 = 16
:
: ***** 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 6400 events for training and 1600 for testing
: Compute initial loss on the validation data
: Training phase 1 of 1: Optimizer ADAM Learning rate = 0.001 regularization 0 minimum error = inf
: --------------------------------------------------------------
: 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 | 0.748783 0.776719 0.988279 0.0827993 7068.08 0
: 2 | 0.576269 0.848801 0.988782 0.0816182 7054.95 1
: 3 Minimum Test error found - save the configuration
: 3 | 0.478644 0.637096 0.998696 0.0837284 6994.79 0
: 4 Minimum Test error found - save the configuration
: 4 | 0.41329 0.58591 1.00254 0.0826951 6957.7 0
: 5 Minimum Test error found - save the configuration
: 5 | 0.376204 0.491052 0.992019 0.0838092 7046.83 0
: 6 Minimum Test error found - save the configuration
: 6 | 0.35734 0.438385 0.97994 0.0828045 7133.81 0
: 7 | 0.320987 0.595953 0.988757 0.0827738 7064.15 1
: 8 | 0.294629 0.454171 0.990283 0.0824548 7049.8 2
: 9 | 0.287831 0.515986 1.00462 0.0826614 6941.72 3
: 10 Minimum Test error found - save the configuration
: 10 | 0.268617 0.437754 0.986957 0.082578 7076.68 0
: 11 | 0.264421 0.465394 0.996727 0.0819308 6996.09 1
: 12 Minimum Test error found - save the configuration
: 12 | 0.244746 0.401096 1.02442 0.0824051 6793.94 0
: 13 | 0.2335 0.419554 1.05873 0.0825301 6556.04 1
: 14 | 0.205957 0.504197 1.07585 0.0815914 6436.97 2
: 15 | 0.209297 0.59716 1.05242 0.0849379 6615.14 3
: 16 | 0.203502 0.558988 1.05004 0.0981179 6723.22 4
: 17 | 0.206452 0.68994 1.03715 0.0822104 6702 5
: 18 | 0.180122 0.542895 1.05474 0.0823246 6581.55 6
:
: Elapsed time for training with 8000 events: 18.4 sec
: Evaluate deep neural network on CPU using batches with size = 100
:
TMVA_DNN_CPU : [dataset] : Evaluation of TMVA_DNN_CPU on training sample (8000 events)
: Elapsed time for evaluation of 8000 events: 0.416 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 = 16
:
: ***** 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 6400 events for training and 1600 for testing
: Compute initial loss on the validation data
: Training phase 1 of 1: Optimizer ADAM Learning rate = 0.001 regularization 0 minimum error = inf
: --------------------------------------------------------------
: 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 | 0.986379 0.678332 7.1635 0.539298 966.155 0
: 2 Minimum Test error found - save the configuration
: 2 | 0.646851 0.635468 7.08144 0.515128 974.672 0
: 3 Minimum Test error found - save the configuration
: 3 | 0.591384 0.594111 7.06026 0.52273 978.963 0
: 4 Minimum Test error found - save the configuration
: 4 | 0.541796 0.537508 7.14463 0.528003 967.26 0
: 5 Minimum Test error found - save the configuration
: 5 | 0.509069 0.493752 7.08806 0.542849 977.814 0
: 6 Minimum Test error found - save the configuration
: 6 | 0.436962 0.422728 7.12359 0.544925 972.842 0
: 7 Minimum Test error found - save the configuration
: 7 | 0.425959 0.400201 7.15882 0.525655 964.849 0
: 8 | 0.39646 0.413087 7.16375 0.524435 963.954 1
: 9 Minimum Test error found - save the configuration
: 9 | 0.386383 0.393087 7.19644 0.541868 961.745 0
: 10 Minimum Test error found - save the configuration
: 10 | 0.361452 0.374568 7.27911 0.537112 949.273 0
: 11 | 0.407606 0.435508 7.37425 0.524705 934.369 1
: 12 | 0.376483 0.397858 7.9295 0.521818 863.967 2
: 13 | 0.345486 0.415415 8.28979 0.533835 825.173 3
: 14 Minimum Test error found - save the configuration
: 14 | 0.342097 0.368685 8.23582 0.551723 832.889 0
: 15 | 0.339892 0.451846 8.08208 0.563782 851.256 1
: 16 | 0.335397 0.427432 8.13928 0.522979 840.303 2
: 17 Minimum Test error found - save the configuration
: 17 | 0.315325 0.353934 8.15202 0.523742 838.983 0
: 18 | 0.318112 0.375789 7.94728 0.537611 863.736 1
: 19 | 0.330806 0.405558 8.11554 0.545851 845.478 2
: 20 | 0.325954 0.402535 8.05144 0.535705 851.547 3
:
: Elapsed time for training with 8000 events: 152 sec
: Evaluate deep neural network on CPU using batches with size = 100
:
TMVA_CNN_CPU : [dataset] : Evaluation of TMVA_CNN_CPU on training sample (8000 events)
: Elapsed time for evaluation of 8000 events: 2.76 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 : 1.147e-02
: 2 : vars : 1.033e-02
: 3 : vars : 1.017e-02
: 4 : vars : 1.001e-02
: 5 : vars : 9.973e-03
: 6 : vars : 9.706e-03
: 7 : vars : 9.517e-03
: 8 : vars : 9.472e-03
: 9 : vars : 9.431e-03
: 10 : vars : 9.184e-03
: 11 : vars : 9.067e-03
: 12 : vars : 8.955e-03
: 13 : vars : 8.892e-03
: 14 : vars : 8.824e-03
: 15 : vars : 8.791e-03
: 16 : vars : 8.682e-03
: 17 : vars : 8.658e-03
: 18 : vars : 8.597e-03
: 19 : vars : 8.574e-03
: 20 : vars : 8.529e-03
: 21 : vars : 8.490e-03
: 22 : vars : 8.433e-03
: 23 : vars : 8.323e-03
: 24 : vars : 8.321e-03
: 25 : vars : 8.278e-03
: 26 : vars : 7.892e-03
: 27 : vars : 7.878e-03
: 28 : vars : 7.782e-03
: 29 : vars : 7.756e-03
: 30 : vars : 7.608e-03
: 31 : vars : 7.467e-03
: 32 : vars : 7.364e-03
: 33 : vars : 7.294e-03
: 34 : vars : 7.245e-03
: 35 : vars : 7.086e-03
: 36 : vars : 7.035e-03
: 37 : vars : 7.028e-03
: 38 : vars : 6.907e-03
: 39 : vars : 6.762e-03
: 40 : vars : 6.732e-03
: 41 : vars : 6.721e-03
: 42 : vars : 6.674e-03
: 43 : vars : 6.567e-03
: 44 : vars : 6.500e-03
: 45 : vars : 6.409e-03
: 46 : vars : 6.345e-03
: 47 : vars : 6.332e-03
: 48 : vars : 6.283e-03
: 49 : vars : 6.267e-03
: 50 : vars : 6.246e-03
: 51 : vars : 6.234e-03
: 52 : vars : 6.178e-03
: 53 : vars : 6.140e-03
: 54 : vars : 6.122e-03
: 55 : vars : 6.075e-03
: 56 : vars : 6.004e-03
: 57 : vars : 5.996e-03
: 58 : vars : 5.934e-03
: 59 : vars : 5.904e-03
: 60 : vars : 5.875e-03
: 61 : vars : 5.821e-03
: 62 : vars : 5.715e-03
: 63 : vars : 5.688e-03
: 64 : vars : 5.663e-03
: 65 : vars : 5.625e-03
: 66 : vars : 5.591e-03
: 67 : vars : 5.524e-03
: 68 : vars : 5.513e-03
: 69 : vars : 5.509e-03
: 70 : vars : 5.432e-03
: 71 : vars : 5.403e-03
: 72 : vars : 5.383e-03
: 73 : vars : 5.364e-03
: 74 : vars : 5.355e-03
: 75 : vars : 5.323e-03
: 76 : vars : 5.317e-03
: 77 : vars : 5.157e-03
: 78 : vars : 5.154e-03
: 79 : vars : 5.093e-03
: 80 : vars : 5.062e-03
: 81 : vars : 5.055e-03
: 82 : vars : 5.039e-03
: 83 : vars : 4.970e-03
: 84 : vars : 4.868e-03
: 85 : vars : 4.854e-03
: 86 : vars : 4.808e-03
: 87 : vars : 4.802e-03
: 88 : vars : 4.694e-03
: 89 : vars : 4.635e-03
: 90 : vars : 4.610e-03
: 91 : vars : 4.602e-03
: 92 : vars : 4.555e-03
: 93 : vars : 4.544e-03
: 94 : vars : 4.540e-03
: 95 : vars : 4.523e-03
: 96 : vars : 4.475e-03
: 97 : vars : 4.454e-03
: 98 : vars : 4.449e-03
: 99 : vars : 4.434e-03
: 100 : vars : 4.417e-03
: 101 : vars : 4.416e-03
: 102 : vars : 4.351e-03
: 103 : vars : 4.329e-03
: 104 : vars : 4.310e-03
: 105 : vars : 4.305e-03
: 106 : vars : 4.250e-03
: 107 : vars : 4.224e-03
: 108 : vars : 4.204e-03
: 109 : vars : 4.143e-03
: 110 : vars : 4.074e-03
: 111 : vars : 4.071e-03
: 112 : vars : 4.051e-03
: 113 : vars : 4.039e-03
: 114 : vars : 4.011e-03
: 115 : vars : 3.985e-03
: 116 : vars : 3.970e-03
: 117 : vars : 3.970e-03
: 118 : vars : 3.965e-03
: 119 : vars : 3.941e-03
: 120 : vars : 3.925e-03
: 121 : vars : 3.907e-03
: 122 : vars : 3.834e-03
: 123 : vars : 3.821e-03
: 124 : vars : 3.818e-03
: 125 : vars : 3.760e-03
: 126 : vars : 3.739e-03
: 127 : vars : 3.723e-03
: 128 : vars : 3.669e-03
: 129 : vars : 3.652e-03
: 130 : vars : 3.630e-03
: 131 : vars : 3.612e-03
: 132 : vars : 3.559e-03
: 133 : vars : 3.513e-03
: 134 : vars : 3.509e-03
: 135 : vars : 3.475e-03
: 136 : vars : 3.475e-03
: 137 : vars : 3.438e-03
: 138 : vars : 3.412e-03
: 139 : vars : 3.369e-03
: 140 : vars : 3.366e-03
: 141 : vars : 3.271e-03
: 142 : vars : 3.264e-03
: 143 : vars : 3.183e-03
: 144 : vars : 3.139e-03
: 145 : vars : 3.110e-03
: 146 : vars : 3.108e-03
: 147 : vars : 2.979e-03
: 148 : vars : 2.969e-03
: 149 : vars : 2.938e-03
: 150 : vars : 2.935e-03
: 151 : vars : 2.900e-03
: 152 : vars : 2.892e-03
: 153 : vars : 2.839e-03
: 154 : vars : 2.838e-03
: 155 : vars : 2.831e-03
: 156 : vars : 2.829e-03
: 157 : vars : 2.785e-03
: 158 : vars : 2.785e-03
: 159 : vars : 2.764e-03
: 160 : vars : 2.753e-03
: 161 : vars : 2.680e-03
: 162 : vars : 2.653e-03
: 163 : vars : 2.598e-03
: 164 : vars : 2.587e-03
: 165 : vars : 2.585e-03
: 166 : vars : 2.575e-03
: 167 : vars : 2.559e-03
: 168 : vars : 2.530e-03
: 169 : vars : 2.457e-03
: 170 : vars : 2.436e-03
: 171 : vars : 2.427e-03
: 172 : vars : 2.425e-03
: 173 : vars : 2.422e-03
: 174 : vars : 2.411e-03
: 175 : vars : 2.408e-03
: 176 : vars : 2.401e-03
: 177 : vars : 2.367e-03
: 178 : vars : 2.364e-03
: 179 : vars : 2.344e-03
: 180 : vars : 2.306e-03
: 181 : vars : 2.294e-03
: 182 : vars : 2.268e-03
: 183 : vars : 2.241e-03
: 184 : vars : 2.217e-03
: 185 : vars : 2.208e-03
: 186 : vars : 2.199e-03
: 187 : vars : 2.111e-03
: 188 : vars : 2.103e-03
: 189 : vars : 2.079e-03
: 190 : vars : 2.062e-03
: 191 : vars : 2.031e-03
: 192 : vars : 2.021e-03
: 193 : vars : 1.940e-03
: 194 : vars : 1.935e-03
: 195 : vars : 1.882e-03
: 196 : vars : 1.872e-03
: 197 : vars : 1.868e-03
: 198 : vars : 1.866e-03
: 199 : vars : 1.828e-03
: 200 : vars : 1.818e-03
: 201 : vars : 1.795e-03
: 202 : vars : 1.789e-03
: 203 : vars : 1.777e-03
: 204 : vars : 1.761e-03
: 205 : vars : 1.730e-03
: 206 : vars : 1.727e-03
: 207 : vars : 1.628e-03
: 208 : vars : 1.594e-03
: 209 : vars : 1.562e-03
: 210 : vars : 1.539e-03
: 211 : vars : 1.530e-03
: 212 : vars : 1.451e-03
: 213 : vars : 1.414e-03
: 214 : vars : 1.395e-03
: 215 : vars : 1.303e-03
: 216 : vars : 1.297e-03
: 217 : vars : 1.243e-03
: 218 : vars : 1.046e-03
: 219 : vars : 1.041e-03
: 220 : vars : 8.890e-04
: 221 : vars : 8.573e-04
: 222 : vars : 0.000e+00
: 223 : vars : 0.000e+00
: 224 : vars : 0.000e+00
: 225 : vars : 0.000e+00
: 226 : vars : 0.000e+00
: 227 : vars : 0.000e+00
: 228 : vars : 0.000e+00
: 229 : vars : 0.000e+00
: 230 : vars : 0.000e+00
: 231 : vars : 0.000e+00
: 232 : vars : 0.000e+00
: 233 : vars : 0.000e+00
: 234 : vars : 0.000e+00
: 235 : vars : 0.000e+00
: 236 : vars : 0.000e+00
: 237 : vars : 0.000e+00
: 238 : vars : 0.000e+00
: 239 : vars : 0.000e+00
: 240 : vars : 0.000e+00
: 241 : vars : 0.000e+00
: 242 : vars : 0.000e+00
: 243 : vars : 0.000e+00
: 244 : vars : 0.000e+00
: 245 : vars : 0.000e+00
: 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= 5.87059
TH1.Print Name = TrainingHistory_TMVA_DNN_CPU_valError, Entries= 0, Total sum= 9.96105
TH1.Print Name = TrainingHistory_TMVA_CNN_CPU_trainingError, Entries= 0, Total sum= 8.71985
TH1.Print Name = TrainingHistory_TMVA_CNN_CPU_valError, Entries= 0, Total sum= 8.9774
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 (2000 events)
: Elapsed time for evaluation of 2000 events: 0.0407 sec
Factory : Test method: TMVA_DNN_CPU for Classification performance
:
: Evaluate deep neural network on CPU using batches with size = 1000
:
TMVA_DNN_CPU : [dataset] : Evaluation of TMVA_DNN_CPU on testing sample (2000 events)
: Elapsed time for evaluation of 2000 events: 0.0924 sec
Factory : Test method: TMVA_CNN_CPU for Classification performance
:
: Evaluate deep neural network on CPU using batches with size = 1000
:
TMVA_CNN_CPU : [dataset] : Evaluation of TMVA_CNN_CPU on testing sample (2000 events)
: Elapsed time for evaluation of 2000 events: 0.793 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 : 0.910
: dataset TMVA_DNN_CPU : 0.890
: dataset BDT : 0.817
: -------------------------------------------------------------------------------------------------------------------
:
: 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 : 0.355 (0.423) 0.722 (0.791) 0.915 (0.937)
: dataset TMVA_DNN_CPU : 0.215 (0.525) 0.668 (0.823) 0.888 (0.944)
: dataset BDT : 0.155 (0.273) 0.529 (0.662) 0.771 (0.876)
: -------------------------------------------------------------------------------------------------------------------
:
Dataset:dataset : Created tree 'TestTree' with 2000 events
:
Dataset:dataset : Created tree 'TrainTree' with 8000 events
:
Factory : ␛[1mThank you for using TMVA!␛[0m
: ␛[1mFor citation information, please visit: http://tmva.sf.net/citeTMVA.html␛[0m
void MakeImagesTree(
int n,
int nh,
int nw)
{
const int ntot = nh * nw;
const int nRndmEvts = 10000;
double delta_sigma = 0.1;
double pixelNoise = 5;
double sX1 = 3;
double sY1 = 3;
double sX2 = sX1 + delta_sigma;
double sY2 = sY1 - delta_sigma;
auto h1 =
new TH2D(
"h1",
"h1", nh, 0, 10, nw, 0, 10);
auto h2 =
new TH2D(
"h2",
"h2", nh, 0, 10, nw, 0, 10);
auto f1 =
new TF2(
"f1",
"xygaus");
auto f2 =
new TF2(
"f2",
"xygaus");
TTree sgn(
"sig_tree",
"signal_tree");
TTree bkg(
"bkg_tree",
"bakground_tree");
TFile f(fileOutName,
"RECREATE");
std::vector<float>
x1(ntot);
std::vector<float>
x2(ntot);
std::vector<float> *px1 = &
x1;
std::vector<float> *px2 = &
x2;
bkg.Branch("vars", "std::vector<float>", &px1);
sgn.Branch("vars", "std::vector<float>", &px2);
f2->SetParameters(1, 5, sX2, 5, sY2);
std::cout << "Filling ROOT tree " << std::endl;
for (
int i = 0; i <
n; ++i) {
if (i % 1000 == 0)
std::cout << "Generating image event ... " << i << std::endl;
h2->Reset();
h2->FillRandom("f2", nRndmEvts);
for (int k = 0; k < nh; ++k) {
for (
int l = 0;
l < nw; ++
l) {
}
}
sgn.Fill();
bkg.Fill();
}
sgn.Write();
bkg.Write();
Info(
"MakeImagesTree",
"Signal and background tree with images data written to the file %s",
f.GetName());
sgn.Print();
bkg.Print();
}
void TMVA_CNN_Classification(std::vector<bool> opt = {1, 1, 1, 1})
{
bool useTMVACNN = (opt.size() > 0) ? opt[0] : false;
bool useKerasCNN = (opt.size() > 1) ? opt[1] : false;
bool useTMVADNN = (opt.size() > 2) ? opt[2] : false;
bool useTMVABDT = (opt.size() > 3) ? opt[3] : false;
#ifndef R__HAS_TMVACPU
#ifndef R__HAS_TMVAGPU
"TMVA is not build with GPU or CPU multi-thread support. Cannot use TMVA Deep Learning for CNN");
useTMVACNN = false;
#endif
#endif
bool writeOutputFile = true;
int num_threads = 0;
if (num_threads >= 0) {
}
else
#ifdef R__HAS_PYMVA
#else
useKerasCNN = false;
#endif
TFile *outputFile =
nullptr;
if (writeOutputFile)
outputFile =
TFile::Open(
"TMVA_CNN_ClassificationOutput.root",
"RECREATE");
"TMVA_CNN_Classification", outputFile,
"!V:ROC:!Silent:Color:AnalysisType=Classification:Transformations=None:!Correlations");
int imgSize = 16 * 16;
TString inputFileName =
"images_data_16x16.root";
if (!fileExist) {
MakeImagesTree(5000, 16, 16);
}
if (!inputFile) {
Error(
"TMVA_CNN_Classification",
"Error opening input file %s - exit", inputFileName.
Data());
return;
}
TTree *signalTree = (
TTree *)inputFile->Get(
"sig_tree");
TTree *backgroundTree = (
TTree *)inputFile->Get(
"bkg_tree");
int nTrainSig = 0.8 * nEventsSig;
int nTrainBkg = 0.8 * nEventsBkg;
"nTrain_Signal=%d:nTrain_Background=%d:SplitMode=Random:SplitSeed=100:NormMode=NumEvents:!V:!CalcCorrelations",
nTrainSig, nTrainBkg);
if (useTMVABDT) {
"!V:NTrees=400:MinNodeSize=2.5%:MaxDepth=2:BoostType=AdaBoost:AdaBoostBeta=0.5:"
"UseBaggedBoost:BaggedSampleFraction=0.5:SeparationType=GiniIndex:nCuts=20");
}
if (useTMVADNN) {
"Layout=DENSE|100|RELU,BNORM,DENSE|100|RELU,BNORM,DENSE|100|RELU,BNORM,DENSE|100|RELU,DENSE|1|LINEAR");
TString trainingString1(
"LearningRate=1e-3,Momentum=0.9,Repetitions=1,"
"ConvergenceSteps=5,BatchSize=100,TestRepetitions=1,"
"MaxEpochs=20,WeightDecay=1e-4,Regularization=None,"
"Optimizer=ADAM,DropConfig=0.0+0.0+0.0+0.");
TString trainingStrategyString(
"TrainingStrategy=");
trainingStrategyString += trainingString1;
TString dnnOptions(
"!H:V:ErrorStrategy=CROSSENTROPY:VarTransform=None:"
"WeightInitialization=XAVIER");
dnnOptions.Append(":");
dnnOptions.Append(layoutString);
dnnOptions.Append(":");
dnnOptions.Append(trainingStrategyString);
TString dnnMethodName =
"TMVA_DNN_CPU";
#ifdef R__HAS_TMVAGPU
dnnOptions += ":Architecture=GPU";
dnnMethodName = "TMVA_DNN_GPU";
#elif defined(R__HAS_TMVACPU)
dnnOptions += ":Architecture=CPU";
#endif
}
if (useTMVACNN) {
TString inputLayoutString(
"InputLayout=1|16|16");
TString layoutString(
"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");
TString trainingString1(
"LearningRate=1e-3,Momentum=0.9,Repetitions=1,"
"ConvergenceSteps=5,BatchSize=100,TestRepetitions=1,"
"MaxEpochs=20,WeightDecay=1e-4,Regularization=None,"
"Optimizer=ADAM,DropConfig=0.0+0.0+0.0+0.0");
TString trainingStrategyString(
"TrainingStrategy=");
trainingStrategyString +=
trainingString1;
TString cnnOptions(
"!H:V:ErrorStrategy=CROSSENTROPY:VarTransform=None:"
"WeightInitialization=XAVIER");
cnnOptions.Append(":");
cnnOptions.Append(inputLayoutString);
cnnOptions.Append(":");
cnnOptions.Append(layoutString);
cnnOptions.Append(":");
cnnOptions.Append(trainingStrategyString);
TString cnnMethodName =
"TMVA_CNN_CPU";
#ifdef R__HAS_TMVAGPU
cnnOptions += ":Architecture=GPU";
cnnMethodName = "TMVA_CNN_GPU";
#else
cnnOptions += ":Architecture=CPU";
cnnMethodName = "TMVA_CNN_CPU";
#endif
}
if (useKerasCNN) {
Info(
"TMVA_CNN_Classification",
"Building convolutional keras model");
m.AddLine(
"import keras");
m.AddLine(
"from keras.models import Sequential");
m.AddLine(
"from keras.optimizers import Adam");
"from keras.layers import Input, Dense, Dropout, Flatten, Conv2D, MaxPooling2D, Reshape, BatchNormalization");
m.AddLine(
"model = keras.models.Sequential() ");
m.AddLine(
"model.add(Reshape((16, 16, 1), input_shape = (256, )))");
m.AddLine(
"model.add(Conv2D(10, kernel_size = (3, 3), kernel_initializer = 'glorot_normal',activation = "
"'relu', padding = 'same'))");
m.AddLine(
"model.add(BatchNormalization())");
m.AddLine(
"model.add(Conv2D(10, kernel_size = (3, 3), kernel_initializer = 'glorot_normal',activation = "
"'relu', padding = 'same'))");
m.AddLine(
"model.add(MaxPooling2D(pool_size = (2, 2), strides = (1,1))) ");
m.AddLine(
"model.add(Flatten())");
m.AddLine(
"model.add(Dense(256, activation = 'relu')) ");
m.AddLine(
"model.add(Dense(2, activation = 'sigmoid')) ");
m.AddLine(
"model.compile(loss = 'binary_crossentropy', optimizer = Adam(lr = 0.001), metrics = ['accuracy'])");
m.AddLine(
"model.save('model_cnn.h5')");
m.AddLine(
"model.summary()");
m.SaveSource(
"make_cnn_model.py");
Warning(
"TMVA_CNN_Classification",
"Error creating Keras model file - skip using Keras");
} else {
Info(
"TMVA_CNN_Classification",
"Booking Keras CNN model");
factory.BookMethod(
"H:!V:VarTransform=None:FilenameModel=model_cnn.h5:"
"FilenameTrainedModel=trained_model_cnn.h5:NumEpochs=20:BatchSize=100:"
"GpuOptions=allow_growth=True");
}
}
factory.TrainAllMethods();
factory.TestAllMethods();
factory.EvaluateAllMethods();
auto c1 = factory.GetROCCurve(loader);
}
static const double x2[5]
static const double x1[5]
void Info(const char *location, const char *msgfmt,...)
void Error(const char *location, const char *msgfmt,...)
void Warning(const char *location, const char *msgfmt,...)
R__EXTERN TRandom * gRandom
R__EXTERN TSystem * gSystem
A specialized string object used for TTree selections.
virtual void SetParameters(const Double_t *params)
virtual void SetParameter(Int_t param, Double_t value)
A 2-Dim function with parameters.
A ROOT file is a suite of consecutive data records (TKey instances) with a well defined format.
static TFile * Open(const char *name, Option_t *option="", const char *ftitle="", Int_t compress=ROOT::RCompressionSetting::EDefaults::kUseCompiledDefault, Int_t netopt=0)
Create / open a file.
void Close(Option_t *option="") override
Close a file.
virtual void Reset(Option_t *option="")
Reset.
virtual void FillRandom(const char *fname, Int_t ntimes=5000)
Fill histogram following distribution in function fname.
virtual Double_t GetBinContent(Int_t bin) const
Return content of bin number bin.
2-D histogram with a double per channel (see TH1 documentation)}
void AddVariablesArray(const TString &expression, int size, char type='F', Double_t min=0, Double_t max=0)
user inserts discriminating array of variables in data set info in case input tree provides an array ...
void AddSignalTree(TTree *signal, Double_t weight=1.0, Types::ETreeType treetype=Types::kMaxTreeType)
number of signal events (used to compute significance)
void PrepareTrainingAndTestTree(const TCut &cut, const TString &splitOpt)
prepare the training and test trees -> same cuts for signal and background
void AddBackgroundTree(TTree *background, Double_t weight=1.0, Types::ETreeType treetype=Types::kMaxTreeType)
number of signal events (used to compute significance)
This is the main MVA steering class.
static void PyInitialize()
Initialize Python interpreter.
Class supporting a collection of lines with C++ code.
virtual Double_t Gaus(Double_t mean=0, Double_t sigma=1)
Samples a random number from the standard Normal (Gaussian) Distribution with the given mean and sigm...
virtual void SetSeed(ULong_t seed=0)
Set the random generator seed.
virtual Double_t Uniform(Double_t x1=1)
Returns a uniform deviate on the interval (0, x1).
const char * Data() const
static TString Format(const char *fmt,...)
Static method which formats a string using a printf style format descriptor and return a TString.
virtual Int_t Exec(const char *shellcmd)
Execute a command.
virtual Bool_t AccessPathName(const char *path, EAccessMode mode=kFileExists)
Returns FALSE if one can access a file using the specified access mode.
virtual void Setenv(const char *name, const char *value)
Set environment variable.
A TTree represents a columnar dataset.
virtual Long64_t GetEntries() const
void EnableImplicitMT(UInt_t numthreads=0)
Enable ROOT's implicit multi-threading for all objects and methods that provide an internal paralleli...
UInt_t GetThreadPoolSize()
Returns the size of ROOT's thread pool.