TMVA_RNN_Classification Namespace Reference

Functions
	MakeTimeData (n, ntime, ndim)
	Helper function to generate the time data set make some time data but not of fixed length.

Variables
	Architecture
str	archString = "GPU" if useGPU else "CPU"
	background = inputFile.Get("bkg")
	BaggedSampleFraction
	BatchSize
int	batchSize = 100
	BoostType
	c1 = factory.GetROCCurve(dataloader)
	Train all methods.
	CalcCorrelations
	datainfo = dataloader.GetDataSetInfo()
	add variables - use new AddVariablesArray function
	dataloader = TMVA.DataLoader("dataset")
str	dnnName = "TMVA_DNN"
	ErrorStrategy
	factory
	Declare Factory.
	fileDoesNotExist = ROOT.gSystem.AccessPathName(inputFileName)
	FilenameModel
	FilenameTrainedModel
	H
	Book TMVA BDT.
	inputFile = TFile.Open(inputFileName)
str	inputFileName = "time_data_t10_d30.root"
	InputLayout
	Layout
	loss
	MaxDepth
int	maxepochs = 10
	MinNodeSize
	model = Sequential()
str	modelName = "model_" + rnn_types[i] + ".keras"
	Book Keras recurrent models.
str	mycutb = ""
str	mycuts = ""
	nCuts
int	ninput = 30
	NormMode
int	ntime = 10
int	nTotEvts = 2000
	nTrain_Background
	nTrain_Signal
float	nTrainBkg = 0.8 * nTotEvts
float	nTrainSig = 0.8 * nTotEvts
	NTrees
int	num_threads = 4
	NumEpochs
int	nvar = ninput * ntime
	optimizer
str	outfileName = "data_RNN_" + archString + ".root"
	outputFile = None
	RandomSeed
str	rnn_type = "RNN"
	Book TMVA recurrent models.
list	rnn_types = ["RNN", "LSTM", "GRU"]
str	rnnLayout = str(rnn_type) + "|10|" + str(ninput) + "|" + str(ntime) + "|0|1,RESHAPE|FLAT,DENSE|64|TANH,LINEAR"
	Define RNN layer layout it should be LayerType (RNN or LSTM or GRU) | number of units | number of inputs | time steps | remember output (typically no=0 | return full sequence.
str	rnnName = "TMVA_" + str(rnn_type)
	define the inputlayout string for RNN the input data should be organize as following: / input layout for RNN: time x ndim add after RNN a reshape layer (needed top flatten the output) and a dense layer with 64 units and a last one Note the last layer is linear because when using Crossentropy a Sigmoid is applied already Define the full RNN Noption string adding the final options for all network
	Shrinkage
	signalTree = inputFile.Get("sgn")
	SplitMode
	SplitSeed
	TFile = ROOT.TFile
	TMVA = ROOT.TMVA
str	trainedModelName = "trained_" + modelName
	TrainingStrategy
str	trainingString1 = "LearningRate=1e-3,Momentum=0.0,Repetitions=1,ConvergenceSteps=5,BatchSize=" + str(batchSize)
	Defining Training strategies.
list	use_rnn_type = [1, 1, 1]
int	use_type = 1
	macro for performing a classification using a Recurrent Neural Network
	UseBaggedBoost
bool	useGPU = True
bool	useKeras = False
bool	useTMVA_BDT = False
bool	useTMVA_DNN = True
bool	useTMVA_RNN = True
	V
	ValidationSize
	vars = datainfo.GetListOfVariables()
	VarTransform
	weighted_metrics
	WeightInitialization
bool	writeOutputFile = True

Function Documentation

MakeTimeData()

TMVA_RNN_Classification.MakeTimeData	(		n,
			ntime,
			ndim )

Helper function to generate the time data set make some time data but not of fixed length.

use a poisson with mu = 5 and truncated at 10

Definition at line 48 of file TMVA_RNN_Classification.py.

Variable Documentation

Architecture

TMVA_RNN_Classification.Architecture

Definition at line 347 of file TMVA_RNN_Classification.py.

archString

str TMVA_RNN_Classification.archString = "GPU" if useGPU else "CPU"

Definition at line 188 of file TMVA_RNN_Classification.py.

background

TMVA_RNN_Classification.background = inputFile.Get("bkg")

Definition at line 261 of file TMVA_RNN_Classification.py.

BaggedSampleFraction

TMVA_RNN_Classification.BaggedSampleFraction

Definition at line 464 of file TMVA_RNN_Classification.py.

BatchSize

TMVA_RNN_Classification.BatchSize

Definition at line 440 of file TMVA_RNN_Classification.py.

batchSize

int TMVA_RNN_Classification.batchSize = 100

Definition at line 148 of file TMVA_RNN_Classification.py.

BoostType

TMVA_RNN_Classification.BoostType

Definition at line 461 of file TMVA_RNN_Classification.py.

c1

TMVA_RNN_Classification.c1 = factory.GetROCCurve(dataloader)

Train all methods.

Definition at line 484 of file TMVA_RNN_Classification.py.

CalcCorrelations

TMVA_RNN_Classification.CalcCorrelations

Definition at line 299 of file TMVA_RNN_Classification.py.

datainfo

TMVA_RNN_Classification.datainfo = dataloader.GetDataSetInfo()

add variables - use new AddVariablesArray function

Definition at line 274 of file TMVA_RNN_Classification.py.

dataloader

TMVA_RNN_Classification.dataloader = TMVA.DataLoader("dataset")

Definition at line 258 of file TMVA_RNN_Classification.py.

dnnName

str TMVA_RNN_Classification.dnnName = "TMVA_DNN"

Definition at line 363 of file TMVA_RNN_Classification.py.

ErrorStrategy

TMVA_RNN_Classification.ErrorStrategy

Definition at line 339 of file TMVA_RNN_Classification.py.

factory

TMVA_RNN_Classification.factory

Initial value:

=  TMVA.Factory(
    "TMVAClassification",
    outputFile,
    V=False,
    Silent=False,
    Color=True,
    DrawProgressBar=True,
    Transformations=None,
    Correlations=False,
    AnalysisType="Classification",
    ModelPersistence=True,
)

Declare Factory.

Definition at line 246 of file TMVA_RNN_Classification.py.

fileDoesNotExist

TMVA_RNN_Classification.fileDoesNotExist = ROOT.gSystem.AccessPathName(inputFileName)

Definition at line 203 of file TMVA_RNN_Classification.py.

FilenameModel

TMVA_RNN_Classification.FilenameModel

Definition at line 437 of file TMVA_RNN_Classification.py.

FilenameTrainedModel

TMVA_RNN_Classification.FilenameTrainedModel

Definition at line 438 of file TMVA_RNN_Classification.py.

H

TMVA_RNN_Classification.H

Book TMVA BDT.

Definition at line 337 of file TMVA_RNN_Classification.py.

inputFile

TMVA_RNN_Classification.inputFile = TFile.Open(inputFileName)

Definition at line 210 of file TMVA_RNN_Classification.py.

inputFileName

str TMVA_RNN_Classification.inputFileName = "time_data_t10_d30.root"

Definition at line 201 of file TMVA_RNN_Classification.py.

InputLayout

TMVA_RNN_Classification.InputLayout

Definition at line 344 of file TMVA_RNN_Classification.py.

Layout

TMVA_RNN_Classification.Layout

Definition at line 345 of file TMVA_RNN_Classification.py.

loss

TMVA_RNN_Classification.loss

Definition at line 419 of file TMVA_RNN_Classification.py.

MaxDepth

TMVA_RNN_Classification.MaxDepth

Definition at line 466 of file TMVA_RNN_Classification.py.

maxepochs

int TMVA_RNN_Classification.maxepochs = 10

Definition at line 149 of file TMVA_RNN_Classification.py.

MinNodeSize

TMVA_RNN_Classification.MinNodeSize

Definition at line 460 of file TMVA_RNN_Classification.py.

model

TMVA_RNN_Classification.model = Sequential()

Definition at line 406 of file TMVA_RNN_Classification.py.

modelName

str TMVA_RNN_Classification.modelName = "model_" + rnn_types[i] + ".keras"

Book Keras recurrent models.

Definition at line 388 of file TMVA_RNN_Classification.py.

mycutb

str TMVA_RNN_Classification.mycutb = ""

Definition at line 287 of file TMVA_RNN_Classification.py.

mycuts

str TMVA_RNN_Classification.mycuts = ""

Definition at line 286 of file TMVA_RNN_Classification.py.

nCuts

TMVA_RNN_Classification.nCuts

Definition at line 465 of file TMVA_RNN_Classification.py.

ninput

int TMVA_RNN_Classification.ninput = 30

Definition at line 146 of file TMVA_RNN_Classification.py.

NormMode

TMVA_RNN_Classification.NormMode

Definition at line 297 of file TMVA_RNN_Classification.py.

ntime

int TMVA_RNN_Classification.ntime = 10

Definition at line 147 of file TMVA_RNN_Classification.py.

nTotEvts

int TMVA_RNN_Classification.nTotEvts = 2000

Definition at line 151 of file TMVA_RNN_Classification.py.

nTrain_Background

TMVA_RNN_Classification.nTrain_Background

Definition at line 294 of file TMVA_RNN_Classification.py.

nTrain_Signal

TMVA_RNN_Classification.nTrain_Signal

Definition at line 293 of file TMVA_RNN_Classification.py.

nTrainBkg

float TMVA_RNN_Classification.nTrainBkg = 0.8 * nTotEvts

Definition at line 283 of file TMVA_RNN_Classification.py.

nTrainSig

float TMVA_RNN_Classification.nTrainSig = 0.8 * nTotEvts

Definition at line 282 of file TMVA_RNN_Classification.py.

NTrees

TMVA_RNN_Classification.NTrees

Definition at line 459 of file TMVA_RNN_Classification.py.

num_threads

int TMVA_RNN_Classification.num_threads = 4

Definition at line 23 of file TMVA_RNN_Classification.py.

NumEpochs

TMVA_RNN_Classification.NumEpochs

Definition at line 439 of file TMVA_RNN_Classification.py.

nvar

int TMVA_RNN_Classification.nvar = ninput * ntime

Definition at line 263 of file TMVA_RNN_Classification.py.

optimizer

TMVA_RNN_Classification.optimizer

Definition at line 419 of file TMVA_RNN_Classification.py.

outfileName

str TMVA_RNN_Classification.outfileName = "data_RNN_" + archString + ".root"

Definition at line 218 of file TMVA_RNN_Classification.py.

outputFile

TMVA_RNN_Classification.outputFile = None

Definition at line 219 of file TMVA_RNN_Classification.py.

RandomSeed

TMVA_RNN_Classification.RandomSeed

Definition at line 343 of file TMVA_RNN_Classification.py.

rnn_type

list TMVA_RNN_Classification.rnn_type = "RNN"

Book TMVA recurrent models.

Definition at line 192 of file TMVA_RNN_Classification.py.

rnn_types

list TMVA_RNN_Classification.rnn_types = ["RNN", "LSTM", "GRU"]

Definition at line 170 of file TMVA_RNN_Classification.py.

rnnLayout

str TMVA_RNN_Classification.rnnLayout = str(rnn_type) + "|10|" + str(ninput) + "|" + str(ntime) + "|0|1,RESHAPE|FLAT,DENSE|64|TANH,LINEAR"

Define RNN layer layout it should be LayerType (RNN or LSTM or GRU) | number of units | number of inputs | time steps | remember output (typically no=0 | return full sequence.

Definition at line 319 of file TMVA_RNN_Classification.py.

rnnName

str TMVA_RNN_Classification.rnnName = "TMVA_" + str(rnn_type)

define the inputlayout string for RNN the input data should be organize as following: / input layout for RNN: time x ndim add after RNN a reshape layer (needed top flatten the output) and a dense layer with 64 units and a last one Note the last layer is linear because when using Crossentropy a Sigmoid is applied already Define the full RNN Noption string adding the final options for all network

Definition at line 332 of file TMVA_RNN_Classification.py.

Shrinkage

TMVA_RNN_Classification.Shrinkage

Definition at line 462 of file TMVA_RNN_Classification.py.

signalTree

TMVA_RNN_Classification.signalTree = inputFile.Get("sgn")

Definition at line 260 of file TMVA_RNN_Classification.py.

SplitMode

TMVA_RNN_Classification.SplitMode

Definition at line 295 of file TMVA_RNN_Classification.py.

SplitSeed

TMVA_RNN_Classification.SplitSeed

Definition at line 296 of file TMVA_RNN_Classification.py.

TFile

TMVA_RNN_Classification.TFile = ROOT.TFile

Definition at line 35 of file TMVA_RNN_Classification.py.

TMVA

TMVA_RNN_Classification.TMVA = ROOT.TMVA

Definition at line 34 of file TMVA_RNN_Classification.py.

trainedModelName

str TMVA_RNN_Classification.trainedModelName = "trained_" + modelName

Definition at line 389 of file TMVA_RNN_Classification.py.

TrainingStrategy

TMVA_RNN_Classification.TrainingStrategy

Definition at line 346 of file TMVA_RNN_Classification.py.

trainingString1

TMVA_RNN_Classification.trainingString1 = "LearningRate=1e-3,Momentum=0.0,Repetitions=1,ConvergenceSteps=5,BatchSize=" + str(batchSize)

Defining Training strategies.

Book TMVA fully connected dense layer models.

Different training strings can be concatenate. Use however only one

Definition at line 322 of file TMVA_RNN_Classification.py.

use_rnn_type

list TMVA_RNN_Classification.use_rnn_type = [1, 1, 1]

Definition at line 171 of file TMVA_RNN_Classification.py.

use_type

int TMVA_RNN_Classification.use_type = 1

macro for performing a classification using a Recurrent Neural Network

Parameters

use_type

use_type = 0 use Simple RNN network use_type = 1 use LSTM network use_type = 2 use GRU use_type = 3 build 3 different networks with RNN, LSTM and GRU

Definition at line 145 of file TMVA_RNN_Classification.py.

UseBaggedBoost

TMVA_RNN_Classification.UseBaggedBoost

Definition at line 463 of file TMVA_RNN_Classification.py.

useGPU

str TMVA_RNN_Classification.useGPU = True

Definition at line 177 of file TMVA_RNN_Classification.py.

useKeras

bool TMVA_RNN_Classification.useKeras = False

Definition at line 153 of file TMVA_RNN_Classification.py.

useTMVA_BDT

bool TMVA_RNN_Classification.useTMVA_BDT = False

Definition at line 157 of file TMVA_RNN_Classification.py.

useTMVA_DNN

bool TMVA_RNN_Classification.useTMVA_DNN = True

Definition at line 156 of file TMVA_RNN_Classification.py.

useTMVA_RNN

tuple TMVA_RNN_Classification.useTMVA_RNN = True

Definition at line 155 of file TMVA_RNN_Classification.py.

V

TMVA_RNN_Classification.V

Definition at line 298 of file TMVA_RNN_Classification.py.

ValidationSize

TMVA_RNN_Classification.ValidationSize

Definition at line 342 of file TMVA_RNN_Classification.py.

vars

TMVA_RNN_Classification.vars = datainfo.GetListOfVariables()

Definition at line 275 of file TMVA_RNN_Classification.py.

VarTransform

TMVA_RNN_Classification.VarTransform

Definition at line 340 of file TMVA_RNN_Classification.py.

weighted_metrics

TMVA_RNN_Classification.weighted_metrics

Definition at line 419 of file TMVA_RNN_Classification.py.

WeightInitialization

TMVA_RNN_Classification.WeightInitialization

Definition at line 341 of file TMVA_RNN_Classification.py.

writeOutputFile

bool TMVA_RNN_Classification.writeOutputFile = True

Definition at line 190 of file TMVA_RNN_Classification.py.