Logo ROOT  
Reference Guide
NeuralNet.cxx
Go to the documentation of this file.
1
2
3#include "TMVA/NeuralNet.h"
4
5#include "TMVA/MethodDNN.h"
6
7namespace TMVA
8{
9 namespace DNN
10 {
11
12 std::shared_ptr<std::function<double(double)>> Gauss = std::make_shared<std::function<double(double)>> ([](double value){ const double s = 6.0; return exp (-std::pow(value*s,2.0)); });
13 std::shared_ptr<std::function<double(double)>> GaussComplement = std::make_shared<std::function<double(double)>> ([](double value){ const double s = 6.0; return 1.0 - exp (-std::pow(value*s,2.0)); });
14 std::shared_ptr<std::function<double(double)>> InvGauss = std::make_shared<std::function<double(double)>> ([](double value){ const double s = 6.0; return -2.0 * value * s*s * (*Gauss.get ()) (value); });
15 std::shared_ptr<std::function<double(double)>> InvGaussComplement = std::make_shared<std::function<double(double)>> ([](double value){ const double s = 6.0; return +2.0 * value * s*s * (*GaussComplement.get ()) (value); });
16 std::shared_ptr<std::function<double(double)>> InvLinear = std::make_shared<std::function<double(double)>> ([](double /*value*/){ return 1.0; });
17 std::shared_ptr<std::function<double(double)>> InvReLU = std::make_shared<std::function<double(double)>> ([](double value){ const double margin = 0.0; return value > margin ? 1.0 : 0; });
18 std::shared_ptr<std::function<double(double)>> InvSigmoid = std::make_shared<std::function<double(double)>> ([](double value){ double s = (*Sigmoid.get ()) (value); return s*(1.0-s); });
19 std::shared_ptr<std::function<double(double)>> InvSoftPlus = std::make_shared<std::function<double(double)>> ([](double value){ return 1.0 / (1.0 + std::exp (-value)); });
20 std::shared_ptr<std::function<double(double)>> InvSoftSign = std::make_shared<std::function<double(double)>> ([](double value){ return std::pow ((1.0 - fabs (value)),2.0); });
21 std::shared_ptr<std::function<double(double)>> InvSymmReLU = std::make_shared<std::function<double(double)>> ([](double value){ const double margin = 0.3; return value > margin ? 1.0 : value < -margin ? 1.0 : 0; });
22 std::shared_ptr<std::function<double(double)>> InvTanh = std::make_shared<std::function<double(double)>> ([](double value){ return 1.0 - std::pow (value, 2.0); });
23 std::shared_ptr<std::function<double(double)>> InvTanhShift = std::make_shared<std::function<double(double)>> ([](double value){ return 0.3 + (1.0 - std::pow (value, 2.0)); });
24 std::shared_ptr<std::function<double(double)>> Linear = std::make_shared<std::function<double(double)>> ([](double value){ return value; });
25 std::shared_ptr<std::function<double(double)>> ReLU = std::make_shared<std::function<double(double)>> ([](double value){ const double margin = 0.0; return value > margin ? value-margin : 0; });
26 std::shared_ptr<std::function<double(double)>> Sigmoid = std::make_shared<std::function<double(double)>> ([](double value){ value = std::max (-100.0, std::min (100.0,value)); return 1.0/(1.0 + std::exp (-value)); });
27 std::shared_ptr<std::function<double(double)>> SoftPlus = std::make_shared<std::function<double(double)>> ([](double value){ return std::log (1.0+ std::exp (value)); });
28 std::shared_ptr<std::function<double(double)>> ZeroFnc = std::make_shared<std::function<double(double)>> ([](double /*value*/){ return 0; });
29 std::shared_ptr<std::function<double(double)>> Tanh = std::make_shared<std::function<double(double)>> ([](double value){ return tanh (value); });
30 std::shared_ptr<std::function<double(double)>> SymmReLU = std::make_shared<std::function<double(double)>> ([](double value){ const double margin = 0.3; return value > margin ? value-margin : value < -margin ? value+margin : 0; });
31 std::shared_ptr<std::function<double(double)>> TanhShift = std::make_shared<std::function<double(double)>> ([](double value){ return tanh (value-0.3); });
32 std::shared_ptr<std::function<double(double)>> SoftSign = std::make_shared<std::function<double(double)>> ([](double value){ return value / (1.0 + fabs (value)); });
33
34
35 double gaussDouble (double mean, double sigma)
36 {
37 static std::default_random_engine generator;
38 std::normal_distribution<double> distribution (mean, sigma);
39 return distribution (generator);
40 }
41
42
43 double uniformDouble (double minValue, double maxValue)
44 {
45 static std::default_random_engine generator;
46 std::uniform_real_distribution<double> distribution(minValue, maxValue);
47 return distribution(generator);
48 }
49
50
51
52 int randomInt (int maxValue)
53 {
54 static std::default_random_engine generator;
55 std::uniform_int_distribution<int> distribution(0,maxValue-1);
56 return distribution(generator);
57 }
58
59
60 double studenttDouble (double distributionParameter)
61 {
62 static std::default_random_engine generator;
63 std::student_t_distribution<double> distribution (distributionParameter);
64 return distribution (generator);
65 }
66
67
68 LayerData::LayerData (size_t inputSize)
69 : m_hasDropOut (false)
70 , m_isInputLayer (true)
71 , m_hasWeights (false)
72 , m_hasGradients (false)
73 , m_eModeOutput (ModeOutputValues::DIRECT)
74 {
75 m_size = inputSize;
76 m_deltas.assign (m_size, 0);
77 }
78
79
80
82 : m_hasDropOut (false)
83 , m_isInputLayer (true)
84 , m_hasWeights (false)
85 , m_hasGradients (false)
86 , m_eModeOutput (eModeOutput)
87 {
88 m_itInputBegin = itInputBegin;
89 m_itInputEnd = itInputEnd;
90 m_size = std::distance (itInputBegin, itInputEnd);
91 m_deltas.assign (m_size, 0);
92 }
93
94
95
96
97 LayerData::LayerData (size_t _size,
98 const_iterator_type itWeightBegin,
99 iterator_type itGradientBegin,
100 std::shared_ptr<std::function<double(double)>> _activationFunction,
101 std::shared_ptr<std::function<double(double)>> _inverseActivationFunction,
102 ModeOutputValues eModeOutput)
103 : m_size (_size)
104 , m_hasDropOut (false)
105 , m_itConstWeightBegin (itWeightBegin)
106 , m_itGradientBegin (itGradientBegin)
107 , m_activationFunction (_activationFunction)
108 , m_inverseActivationFunction (_inverseActivationFunction)
109 , m_isInputLayer (false)
110 , m_hasWeights (true)
111 , m_hasGradients (true)
112 , m_eModeOutput (eModeOutput)
113 {
114 m_values.assign (_size, 0);
115 m_deltas.assign (_size, 0);
116 m_valueGradients.assign (_size, 0);
117 }
118
119
120
121
122 LayerData::LayerData (size_t _size, const_iterator_type itWeightBegin,
123 std::shared_ptr<std::function<double(double)>> _activationFunction,
124 ModeOutputValues eModeOutput)
125 : m_size (_size)
126 , m_hasDropOut (false)
127 , m_itConstWeightBegin (itWeightBegin)
128 , m_activationFunction (_activationFunction)
129 , m_inverseActivationFunction ()
130 , m_isInputLayer (false)
131 , m_hasWeights (true)
132 , m_hasGradients (false)
133 , m_eModeOutput (eModeOutput)
134 {
135 m_values.assign (_size, 0);
136 }
137
138
139
141 {
142 container_type probabilitiesContainer;
144 {
145 std::transform (begin (m_values), end (m_values), std::back_inserter (probabilitiesContainer), (*Sigmoid.get ()));
146 }
148 {
149 double sum = 0;
150 probabilitiesContainer = m_values;
151 std::for_each (begin (probabilitiesContainer), end (probabilitiesContainer), [&sum](double& p){ p = std::exp (p); sum += p; });
152 if (sum != 0)
153 std::for_each (begin (probabilitiesContainer), end (probabilitiesContainer), [sum ](double& p){ p /= sum; });
154 }
155 else
156 {
157 probabilitiesContainer.assign (begin (m_values), end (m_values));
158 }
159 return probabilitiesContainer;
160 }
161
162
163
164
165
166 Layer::Layer (size_t _numNodes, EnumFunction _activationFunction, ModeOutputValues eModeOutputValues)
167 : m_numNodes (_numNodes)
168 , m_eModeOutputValues (eModeOutputValues)
169 , m_activationFunctionType (_activationFunction)
170 {
171 for (size_t iNode = 0; iNode < _numNodes; ++iNode)
172 {
173 auto actFnc = Linear;
174 auto invActFnc = InvLinear;
175 switch (_activationFunction)
176 {
178 actFnc = ZeroFnc;
179 invActFnc = ZeroFnc;
180 break;
182 actFnc = Linear;
183 invActFnc = InvLinear;
184 break;
186 actFnc = Tanh;
187 invActFnc = InvTanh;
188 break;
190 actFnc = ReLU;
191 invActFnc = InvReLU;
192 break;
194 actFnc = SymmReLU;
195 invActFnc = InvSymmReLU;
196 break;
198 actFnc = TanhShift;
199 invActFnc = InvTanhShift;
200 break;
202 actFnc = SoftSign;
203 invActFnc = InvSoftSign;
204 break;
206 actFnc = Sigmoid;
207 invActFnc = InvSigmoid;
208 break;
210 actFnc = Gauss;
211 invActFnc = InvGauss;
212 break;
214 actFnc = GaussComplement;
215 invActFnc = InvGaussComplement;
216 break;
217 }
218 m_activationFunction = actFnc;
219 m_inverseActivationFunction = invActFnc;
220 }
221 }
222
223
224
225
226
227
228
229
230
231
233 size_t _convergenceSteps, size_t _batchSize, size_t _testRepetitions,
234 double _factorWeightDecay, EnumRegularization eRegularization,
235 MinimizerType _eMinimizerType, double _learningRate,
236 double _momentum, int _repetitions, bool _useMultithreading)
237 : m_timer (100, name)
238 , m_minProgress (0)
239 , m_maxProgress (100)
240 , m_convergenceSteps (_convergenceSteps)
241 , m_batchSize (_batchSize)
242 , m_testRepetitions (_testRepetitions)
243 , m_factorWeightDecay (_factorWeightDecay)
244 , count_E (0)
245 , count_dE (0)
246 , count_mb_E (0)
247 , count_mb_dE (0)
248 , m_regularization (eRegularization)
249 , fLearningRate (_learningRate)
250 , fMomentum (_momentum)
251 , fRepetitions (_repetitions)
252 , fMinimizerType (_eMinimizerType)
253 , m_convergenceCount (0)
254 , m_maxConvergenceCount (0)
255 , m_minError (1e10)
256 , m_useMultithreading (_useMultithreading)
257 , fMonitoring (NULL)
258 {
259 }
260
262 {
263 }
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278 /** \brief action to be done when the training cycle is started (e.g. update some monitoring output)
279 *
280 */
282 {
283 if (fMonitoring)
284 {
285 create ("ROC", 100, 0, 1, 100, 0, 1);
286 create ("Significance", 100, 0, 1, 100, 0, 3);
287 create ("OutputSig", 100, 0, 1);
288 create ("OutputBkg", 100, 0, 1);
289 fMonitoring->ProcessEvents ();
290 }
291 }
292
293 /** \brief action to be done when the training cycle is ended (e.g. update some monitoring output)
294 *
295 */
297 {
298 if (fMonitoring) fMonitoring->ProcessEvents ();
299 }
300
301 /** \brief action to be done after the computation of a test sample (e.g. update some monitoring output)
302 *
303 */
304 void ClassificationSettings::testSample (double /*error*/, double output, double target, double weight)
305 {
306
307 m_output.push_back (output);
308 m_targets.push_back (target);
309 m_weights.push_back (weight);
310 }
311
312
313 /** \brief action to be done when the test cycle is started (e.g. update some monitoring output)
314 *
315 */
317 {
318 m_output.clear ();
319 m_targets.clear ();
320 m_weights.clear ();
321 }
322
323 /** \brief action to be done when the training cycle is ended (e.g. update some monitoring output)
324 *
325 */
327 {
328 if (m_output.empty ())
329 return;
330 double minVal = *std::min_element (begin (m_output), end (m_output));
331 double maxVal = *std::max_element (begin (m_output), end (m_output));
332 const size_t numBinsROC = 1000;
333 const size_t numBinsData = 100;
334
335 std::vector<double> truePositives (numBinsROC+1, 0);
336 std::vector<double> falsePositives (numBinsROC+1, 0);
337 std::vector<double> trueNegatives (numBinsROC+1, 0);
338 std::vector<double> falseNegatives (numBinsROC+1, 0);
339
340 std::vector<double> x (numBinsData, 0);
341 std::vector<double> datSig (numBinsData+1, 0);
342 std::vector<double> datBkg (numBinsData+1, 0);
343
344 double binSizeROC = (maxVal - minVal)/(double)numBinsROC;
345 double binSizeData = (maxVal - minVal)/(double)numBinsData;
346
347 double sumWeightsSig = 0.0;
348 double sumWeightsBkg = 0.0;
349
350 for (size_t b = 0; b < numBinsData; ++b)
351 {
352 double binData = minVal + b*binSizeData;
353 x.at (b) = binData;
354 }
355
356 if (fabs(binSizeROC) < 0.0001)
357 return;
358
359 for (size_t i = 0, iEnd = m_output.size (); i < iEnd; ++i)
360 {
361 double val = m_output.at (i);
362 double truth = m_targets.at (i);
363 double weight = m_weights.at (i);
364
365 bool isSignal = (truth > 0.5 ? true : false);
366
367 if (m_sumOfSigWeights != 0 && m_sumOfBkgWeights != 0)
368 {
369 if (isSignal)
370 weight *= m_sumOfSigWeights;
371 else
372 weight *= m_sumOfBkgWeights;
373 }
374
375 size_t binROC = (val-minVal)/binSizeROC;
376 size_t binData = (val-minVal)/binSizeData;
377
378 if (isSignal)
379 {
380 for (size_t n = 0; n <= binROC; ++n)
381 {
382 truePositives.at (n) += weight;
383 }
384 for (size_t n = binROC+1; n < numBinsROC; ++n)
385 {
386 falseNegatives.at (n) += weight;
387 }
388
389 datSig.at (binData) += weight;
390 sumWeightsSig += weight;
391 }
392 else
393 {
394 for (size_t n = 0; n <= binROC; ++n)
395 {
396 falsePositives.at (n) += weight;
397 }
398 for (size_t n = binROC+1; n < numBinsROC; ++n)
399 {
400 trueNegatives.at (n) += weight;
401 }
402
403 datBkg.at (binData) += weight;
404 sumWeightsBkg += weight;
405 }
406 }
407
408 std::vector<double> sigEff;
409 std::vector<double> backRej;
410
411 double bestSignificance = 0;
412 double bestCutSignificance = 0;
413
414 double numEventsScaleFactor = 1.0;
415 if (m_scaleToNumEvents > 0)
416 {
417 size_t numEvents = m_output.size ();
418 numEventsScaleFactor = double (m_scaleToNumEvents)/double (numEvents);
419 }
420
421 clear ("ROC");
422 clear ("Significance");
423
424 for (size_t i = 0; i < numBinsROC; ++i)
425 {
426 double tp = truePositives.at (i) * numEventsScaleFactor;
427 double fp = falsePositives.at (i) * numEventsScaleFactor;
428 double tn = trueNegatives.at (i) * numEventsScaleFactor;
429 double fn = falseNegatives.at (i) * numEventsScaleFactor;
430
431 double seff = (tp+fn == 0.0 ? 1.0 : (tp / (tp+fn)));
432 double brej = (tn+fp == 0.0 ? 0.0 : (tn / (tn+fp)));
433
434 sigEff.push_back (seff);
435 backRej.push_back (brej);
436
437 // m_histROC->Fill (seff, brej);
438 addPoint ("ROC", seff, brej); // x, y
439
440
441 double currentCut = (i * binSizeROC)+minVal;
442
443 double sig = tp;
444 double bkg = fp;
445 double significance = sig / sqrt (sig + bkg);
446 if (significance > bestSignificance)
447 {
448 bestSignificance = significance;
449 bestCutSignificance = currentCut;
450 }
451
452 addPoint ("Significance", currentCut, significance);
453 // m_histSignificance->Fill (currentCut, significance);
454 }
455
456 m_significances.push_back (bestSignificance);
457 static size_t testCycle = 0;
458
459 clear ("OutputSig");
460 clear ("OutputBkg");
461 for (size_t i = 0; i < numBinsData; ++i)
462 {
463 addPoint ("OutputSig", x.at (i), datSig.at (i)/sumWeightsSig);
464 addPoint ("OutputBkg", x.at (i), datBkg.at (i)/sumWeightsBkg);
465 // m_histOutputSignal->Fill (x.at (i), datSig.at (1)/sumWeightsSig);
466 // m_histOutputBackground->Fill (x.at (i), datBkg.at (1)/sumWeightsBkg);
467 }
468
469
470 ++testCycle;
471
472 if (fMonitoring)
473 {
474 plot ("ROC", "", 2, kRed);
475 plot ("Significance", "", 3, kRed);
476 plot ("OutputSig", "", 4, kRed);
477 plot ("OutputBkg", "same", 4, kBlue);
478 fMonitoring->ProcessEvents ();
479 }
480
481 m_cutValue = bestCutSignificance;
482 }
483
484
485 /** \brief check for convergence
486 *
487 */
488 bool Settings::hasConverged (double testError)
489 {
490 // std::cout << "check convergence; minError " << m_minError << " current " << testError
491 // << " current convergence count " << m_convergenceCount << std::endl;
492 if (testError < m_minError*0.999)
493 {
495 m_minError = testError;
496 }
497 else
498 {
501 }
502
503
504 if (m_convergenceCount >= convergenceSteps () || testError <= 0)
505 return true;
506
507 return false;
508 }
509
510
511
512 /** \brief set the weight sums to be scaled to (preparations for monitoring output)
513 *
514 */
515 void ClassificationSettings::setWeightSums (double sumOfSigWeights, double sumOfBkgWeights)
516 {
517 m_sumOfSigWeights = sumOfSigWeights; m_sumOfBkgWeights = sumOfBkgWeights;
518 }
519
520 /** \brief preparation for monitoring output
521 *
522 */
524 std::string _fileNameNetConfig,
525 std::string _fileNameResult,
526 std::vector<Pattern>* _resultPatternContainer)
527 {
528 m_pResultPatternContainer = _resultPatternContainer;
529 m_fileNameResult = _fileNameResult;
530 m_fileNameNetConfig = _fileNameNetConfig;
531 }
532
533
534
535
536
537
538
539
540 /** \brief compute the number of weights given the size of the input layer
541 *
542 */
543 size_t Net::numWeights (size_t trainingStartLayer) const
544 {
545 size_t num (0);
546 size_t index (0);
547 size_t prevNodes (inputSize ());
548 for (auto& layer : m_layers)
549 {
550 if (index >= trainingStartLayer)
551 num += layer.numWeights (prevNodes);
552 prevNodes = layer.numNodes ();
553 ++index;
554 }
555 return num;
556 }
557
558
559 size_t Net::numNodes (size_t trainingStartLayer) const
560 {
561 size_t num (0);
562 size_t index (0);
563 for (auto& layer : m_layers)
564 {
565 if (index >= trainingStartLayer)
566 num += layer.numNodes ();
567 ++index;
568 }
569 return num;
570 }
571
572 /** \brief prepare the drop-out container given the provided drop-fractions
573 *
574 */
575 void Net::fillDropContainer (DropContainer& dropContainer, double dropFraction, size_t _numNodes) const
576 {
577 size_t numDrops = dropFraction * _numNodes;
578 if (numDrops >= _numNodes) // maintain at least one node
579 numDrops = _numNodes - 1;
580 // add the markers for the nodes which are enabled
581 dropContainer.insert (end (dropContainer), _numNodes-numDrops, true);
582 // add the markers for the disabled nodes
583 dropContainer.insert (end (dropContainer), numDrops, false);
584 // shuffle enabled and disabled markers
585 std::shuffle(end(dropContainer)-_numNodes, end(dropContainer), std::default_random_engine{});
586 }
587
588 }; // namespace DNN
589}; // namespace TMVA
590
#define b(i)
Definition: RSha256.hxx:100
@ kRed
Definition: Rtypes.h:64
@ kBlue
Definition: Rtypes.h:64
char name[80]
Definition: TGX11.cxx:109
double tanh(double)
double pow(double, double)
double sqrt(double)
double exp(double)
double log(double)
void startTrainCycle()
action to be done when the training cycle is started (e.g.
Definition: NeuralNet.cxx:281
void endTrainCycle(double)
action to be done when the training cycle is ended (e.g.
Definition: NeuralNet.cxx:296
virtual void endTestCycle()
action to be done when the training cycle is ended (e.g.
Definition: NeuralNet.cxx:326
std::vector< Pattern > * m_pResultPatternContainer
Definition: NeuralNet.h:1016
void setResultComputation(std::string _fileNameNetConfig, std::string _fileNameResult, std::vector< Pattern > *_resultPatternContainer)
preparation for monitoring output
Definition: NeuralNet.cxx:523
std::vector< double > m_significances
Definition: NeuralNet.h:1008
std::vector< double > m_weights
Definition: NeuralNet.h:1005
std::vector< double > m_targets
Definition: NeuralNet.h:1004
void testSample(double error, double output, double target, double weight)
action to be done after the computation of a test sample (e.g.
Definition: NeuralNet.cxx:304
virtual void startTestCycle()
action to be done when the test cycle is started (e.g.
Definition: NeuralNet.cxx:316
void setWeightSums(double sumOfSigWeights, double sumOfBkgWeights)
set the weight sums to be scaled to (preparations for monitoring output)
Definition: NeuralNet.cxx:515
std::vector< double > m_output
Definition: NeuralNet.h:1003
const_iterator_type m_itInputBegin
iterator to the first of the nodes in the input node vector
Definition: NeuralNet.h:643
std::vector< double > m_deltas
stores the deltas for the DNN training
Definition: NeuralNet.h:646
container_type::iterator iterator_type
Definition: NeuralNet.h:442
LayerData(const_iterator_type itInputBegin, const_iterator_type itInputEnd, ModeOutputValues eModeOutput=ModeOutputValues::DIRECT)
c'tor of LayerData
Definition: NeuralNet.cxx:81
std::vector< double > container_type
Definition: NeuralNet.h:440
std::vector< double > m_values
stores the values of the nodes in this layer
Definition: NeuralNet.h:648
const_iterator_type m_itInputEnd
iterator to the end of the nodes in the input node vector
Definition: NeuralNet.h:644
container_type::const_iterator const_iterator_type
Definition: NeuralNet.h:443
std::vector< double > m_valueGradients
stores the gradients of the values (nodes)
Definition: NeuralNet.h:647
container_type computeProbabilities() const
compute the probabilities from the node values
Definition: NeuralNet.cxx:140
ModeOutputValues m_eModeOutput
stores the output mode (DIRECT, SIGMOID, SOFTMAX)
Definition: NeuralNet.h:662
std::shared_ptr< std::function< double(double)> > m_activationFunction
stores the activation function
Definition: NeuralNet.h:703
std::shared_ptr< std::function< double(double)> > m_inverseActivationFunction
stores the inverse activation function
Definition: NeuralNet.h:704
Layer(size_t numNodes, EnumFunction activationFunction, ModeOutputValues eModeOutputValues=ModeOutputValues::DIRECT)
c'tor for defining a Layer
Definition: NeuralNet.cxx:166
std::vector< Layer > m_layers
layer-structure-data
Definition: NeuralNet.h:1279
size_t inputSize() const
input size of the DNN
Definition: NeuralNet.h:1105
size_t numNodes(size_t trainingStartLayer=0) const
returns the number of nodes in this net
Definition: NeuralNet.cxx:559
void fillDropContainer(DropContainer &dropContainer, double dropFraction, size_t numNodes) const
prepare the drop-out-container (select the nodes which are to be dropped out)
Definition: NeuralNet.cxx:575
size_t numWeights(size_t trainingStartLayer=0) const
returns the number of weights in this net
Definition: NeuralNet.cxx:543
void clear(std::string histoName)
for monitoring
Definition: NeuralNet.h:831
virtual bool hasConverged(double testError)
has this training converged already?
Definition: NeuralNet.cxx:488
Settings(TString name, size_t _convergenceSteps=15, size_t _batchSize=10, size_t _testRepetitions=7, double _factorWeightDecay=1e-5, TMVA::DNN::EnumRegularization _regularization=TMVA::DNN::EnumRegularization::NONE, MinimizerType _eMinimizerType=MinimizerType::fSteepest, double _learningRate=1e-5, double _momentum=0.3, int _repetitions=3, bool _multithreading=true)
c'tor
Definition: NeuralNet.cxx:232
void addPoint(std::string histoName, double x)
for monitoring
Definition: NeuralNet.h:828
virtual ~Settings()
d'tor
Definition: NeuralNet.cxx:261
size_t m_convergenceCount
Definition: NeuralNet.h:864
void plot(std::string histoName, std::string options, int pad, EColor color)
for monitoring
Definition: NeuralNet.h:830
size_t convergenceSteps() const
how many steps until training is deemed to have converged
Definition: NeuralNet.h:773
std::shared_ptr< Monitoring > fMonitoring
Definition: NeuralNet.h:872
void create(std::string histoName, int bins, double min, double max)
for monitoring
Definition: NeuralNet.h:826
size_t m_maxConvergenceCount
Definition: NeuralNet.h:865
Basic string class.
Definition: TString.h:131
const Double_t sigma
Double_t x[n]
Definition: legend1.C:17
const Int_t n
Definition: legend1.C:16
VecExpr< UnaryOp< Fabs< T >, VecExpr< A, T, D >, T >, T, D > fabs(const VecExpr< A, T, D > &rhs)
void function(const Char_t *name_, T fun, const Char_t *docstring=0)
Definition: RExports.h:151
static constexpr double s
std::shared_ptr< std::function< double(double)> > InvGauss
Definition: NeuralNet.cxx:14
double uniformDouble(double minValue, double maxValue)
Definition: NeuralNet.cxx:43
std::shared_ptr< std::function< double(double)> > SymmReLU
Definition: NeuralNet.cxx:30
std::shared_ptr< std::function< double(double)> > TanhShift
Definition: NeuralNet.cxx:31
std::shared_ptr< std::function< double(double)> > Tanh
Definition: NeuralNet.cxx:29
std::shared_ptr< std::function< double(double)> > InvSigmoid
Definition: NeuralNet.cxx:18
std::shared_ptr< std::function< double(double)> > SoftPlus
Definition: NeuralNet.cxx:27
EnumRegularization
Definition: NeuralNet.h:174
std::shared_ptr< std::function< double(double)> > ZeroFnc
Definition: NeuralNet.cxx:28
std::shared_ptr< std::function< double(double)> > InvSoftSign
Definition: NeuralNet.cxx:20
double studenttDouble(double distributionParameter)
Definition: NeuralNet.cxx:60
std::shared_ptr< std::function< double(double)> > InvGaussComplement
Definition: NeuralNet.cxx:15
std::shared_ptr< std::function< double(double)> > InvTanh
Definition: NeuralNet.cxx:22
std::shared_ptr< std::function< double(double)> > Linear
Definition: NeuralNet.cxx:24
std::shared_ptr< std::function< double(double)> > InvReLU
Definition: NeuralNet.cxx:17
std::shared_ptr< std::function< double(double)> > GaussComplement
Definition: NeuralNet.cxx:13
std::shared_ptr< std::function< double(double)> > Gauss
Definition: NeuralNet.cxx:12
MinimizerType
< list all the minimizer types
Definition: NeuralNet.h:322
std::shared_ptr< std::function< double(double)> > Sigmoid
Definition: NeuralNet.cxx:26
double gaussDouble(double mean, double sigma)
Definition: NeuralNet.cxx:35
ModeOutputValues
Definition: NeuralNet.h:180
std::shared_ptr< std::function< double(double)> > SoftSign
Definition: NeuralNet.cxx:32
std::shared_ptr< std::function< double(double)> > InvSoftPlus
Definition: NeuralNet.cxx:19
std::shared_ptr< std::function< double(double)> > ReLU
Definition: NeuralNet.cxx:25
bool isFlagSet(T flag, T value)
Definition: NeuralNet.h:213
int randomInt(int maxValue)
Definition: NeuralNet.cxx:52
std::shared_ptr< std::function< double(double)> > InvTanhShift
Definition: NeuralNet.cxx:23
std::vector< char > DropContainer
Definition: NeuralNet.h:220
std::shared_ptr< std::function< double(double)> > InvSymmReLU
Definition: NeuralNet.cxx:21
std::shared_ptr< std::function< double(double)> > InvLinear
Definition: NeuralNet.cxx:16
create variable transformations
static long int sum(long int i)
Definition: Factory.cxx:2276
static void output(int code)
Definition: gifencode.c:226