Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Merge pull request #21215 from mverzett/DeepCSVProbToDiscriminators_C…
…MSSW_9_4_X DeepCSV/DeepCMVA discriminators
- Loading branch information
Showing
9 changed files
with
378 additions
and
10 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
242 changes: 242 additions & 0 deletions
242
RecoBTag/Combined/plugins/BTagProbabilityToDiscriminator.cc
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,242 @@ | ||
| // -*- C++ -*- | ||
| // | ||
| // Package: RecoBTag/SecondaryVertex | ||
| // Class: BTagProbabilityToDiscriminator | ||
| // | ||
| /** | ||
| * | ||
| * Description: EDProducer that performs simple arithmetic on the | ||
| * multi-classifier probabilities to compute simple discriminators | ||
| * | ||
| * Implementation: | ||
| * A collection of output discriminators is defined in a VPSet, each | ||
| * containing the output name, input probabilities and normalization (empty | ||
| * vInputTag if none) the output is computed as | ||
| * sum(INPUTS)/sum(normalizations) | ||
| */ | ||
| // | ||
| // Original Author: Mauro Verzetti (CERN) | ||
| // | ||
| // | ||
|
|
||
| // system include files | ||
| #include <memory> | ||
|
|
||
| // user include files | ||
| #include "FWCore/Framework/interface/Frameworkfwd.h" | ||
| #include "FWCore/Framework/interface/stream/EDProducer.h" | ||
|
|
||
| #include "FWCore/Framework/interface/Event.h" | ||
| #include "FWCore/Framework/interface/MakerMacros.h" | ||
|
|
||
| #include "FWCore/ParameterSet/interface/ParameterSet.h" | ||
| #include "FWCore/Utilities/interface/StreamID.h" | ||
|
|
||
| #include "DataFormats/BTauReco/interface/JetTag.h" | ||
| #include "DataFormats/Common/interface/RefToBase.h" | ||
| #include "FWCore/Framework/interface/makeRefToBaseProdFrom.h" | ||
|
|
||
| // from lwtnn | ||
| #include <fstream> | ||
| #include <iostream> | ||
| #include <map> | ||
| #include <set> | ||
| #include <string> | ||
| #include <vector> | ||
|
|
||
| #include <boost/algorithm/string.hpp> | ||
| #include <unordered_map> | ||
| using namespace std; | ||
| using namespace reco; | ||
| // | ||
| // class declaration | ||
| // | ||
|
|
||
| class BTagProbabilityToDiscriminator : public edm::stream::EDProducer<> { | ||
| public: | ||
| explicit BTagProbabilityToDiscriminator(const edm::ParameterSet &); | ||
| ~BTagProbabilityToDiscriminator() override {} | ||
|
|
||
| static void fillDescriptions(edm::ConfigurationDescriptions &descriptions); | ||
|
|
||
| private: | ||
| typedef std::vector<edm::InputTag> vInputTag; | ||
| typedef std::vector<std::string> vstring; | ||
| typedef std::vector<edm::ParameterSet> vPSet; | ||
| struct Discriminator { | ||
| std::string name; // needed? | ||
| vstring numerator; | ||
| vstring denominator; | ||
| }; | ||
|
|
||
| void beginStream(edm::StreamID) override {} | ||
| void produce(edm::Event &, const edm::EventSetup &) override; | ||
| void endStream() override {} | ||
|
|
||
| // ----------member data --------------------------- | ||
| std::vector<Discriminator> discrims_; | ||
| std::unordered_map<std::string, edm::EDGetTokenT<JetTagCollection>> | ||
| jet_tags_; // caches jet tags to avoid repetitions | ||
| }; | ||
|
|
||
| BTagProbabilityToDiscriminator::BTagProbabilityToDiscriminator( | ||
| const edm::ParameterSet &iConfig) { | ||
| for (auto discriminator : iConfig.getParameter<vPSet>("discriminators")) { | ||
| Discriminator current; | ||
| current.name = discriminator.getParameter<std::string>("name"); | ||
| produces<JetTagCollection>(current.name); | ||
|
|
||
| for (auto intag : discriminator.getParameter<vInputTag>("numerator")) { | ||
| if (jet_tags_.find(intag.encode()) == jet_tags_.end()) { // new | ||
| // probability | ||
| jet_tags_[intag.encode()] = consumes<JetTagCollection>(intag); | ||
| } | ||
| current.numerator.push_back(intag.encode()); | ||
| } | ||
|
|
||
| for (auto intag : discriminator.getParameter<vInputTag>("denominator")) { | ||
| if (jet_tags_.find(intag.encode()) == jet_tags_.end()) { // new | ||
| // probability | ||
| jet_tags_[intag.encode()] = consumes<JetTagCollection>(intag); | ||
| } | ||
| current.denominator.push_back(intag.encode()); | ||
| } | ||
| discrims_.push_back(current); | ||
| } | ||
|
|
||
| if (jet_tags_.empty()) { | ||
| throw cms::Exception("RuntimeError") | ||
| << "The module BTagProbabilityToDiscriminator is run without any input " | ||
| "probability to work on!" | ||
| << std::endl; | ||
| } | ||
| } | ||
|
|
||
| void BTagProbabilityToDiscriminator::produce(edm::Event &iEvent, | ||
| const edm::EventSetup &iSetup) { | ||
| std::unordered_map<std::string, edm::Handle<JetTagCollection>> | ||
| tags; // caches jet tags to avoid repetitions | ||
| size_t size = 0; | ||
| bool first = true; | ||
| for (const auto& entry : jet_tags_) { | ||
| edm::Handle<JetTagCollection> tmp; | ||
| iEvent.getByToken(entry.second, tmp); | ||
| tags[entry.first] = tmp; | ||
| if (first) | ||
| size = tmp->size(); | ||
| else { | ||
| if (tmp->size() != size) { | ||
| throw cms::Exception("RuntimeError") | ||
| << "The length of one of the input jet tag collections does not " | ||
| "match " | ||
| << "with the others, this is probably due to the probabilities " | ||
| "belonging to different jet collections, which is forbidden!" | ||
| << std::endl; | ||
| } | ||
| } | ||
| first = false; | ||
| } | ||
|
|
||
| // create the output collection | ||
| // which is a "map" RefToBase<Jet> --> float | ||
| vector<std::unique_ptr<JetTagCollection>> output_tags; | ||
| output_tags.reserve(discrims_.size()); | ||
| for (size_t i = 0; i < discrims_.size(); ++i) { | ||
| output_tags.push_back(std::make_unique<JetTagCollection>( | ||
| *(tags.begin()->second)) // clone from the first element, will change | ||
| // the content later on | ||
| ); | ||
| } | ||
|
|
||
| // loop over jets | ||
| for (size_t idx = 0; idx < output_tags[0]->size(); idx++) { | ||
| auto key = output_tags[0]->key(idx); // use key only for writing | ||
| // loop over new discriminators to produce | ||
| for (size_t disc_idx = 0; disc_idx < output_tags.size(); disc_idx++) { | ||
| float numerator = 0; | ||
| for (auto &num : discrims_[disc_idx].numerator) | ||
| numerator += (*tags[num])[idx].second; | ||
| float denominator = !discrims_[disc_idx].denominator.empty() ? 0 : 1; | ||
| for (auto &den : discrims_[disc_idx].denominator) | ||
| denominator += (*tags[den])[idx].second; | ||
| //protect against 0 denominator and undefined jet values (numerator probability < 0) | ||
| float new_value = (denominator != 0 && numerator >= 0) ? numerator / denominator : -10.; | ||
| (*output_tags[disc_idx])[key] = new_value; | ||
| } | ||
| } | ||
|
|
||
| // put the output in the event | ||
| for (size_t i = 0; i < output_tags.size(); ++i) { | ||
| iEvent.put(std::move(output_tags[i]), discrims_[i].name); | ||
| } | ||
| } | ||
|
|
||
| // ------------ method fills 'descriptions' with the allowed parameters for the | ||
| // module ------------ | ||
| void BTagProbabilityToDiscriminator::fillDescriptions( | ||
| edm::ConfigurationDescriptions &descriptions) { | ||
| edm::ParameterSetDescription desc; | ||
| { | ||
| edm::ParameterSetDescription vpsd1; | ||
| vpsd1.add<std::vector<edm::InputTag>>("denominator", {}); | ||
| vpsd1.add<std::vector<edm::InputTag>>( | ||
| "numerator", | ||
| { | ||
| edm::InputTag("pfDeepCSVJetTags", "probb"), | ||
| edm::InputTag("pfDeepCSVJetTags", "probbb"), | ||
| }); | ||
| vpsd1.add<std::string>("name", "BvsAll"); | ||
| std::vector<edm::ParameterSet> temp1; | ||
| temp1.reserve(3); | ||
| { | ||
| edm::ParameterSet temp2; | ||
| temp2.addParameter<std::vector<edm::InputTag>>("denominator", {}); | ||
| temp2.addParameter<std::vector<edm::InputTag>>( | ||
| "numerator", | ||
| { | ||
| edm::InputTag("pfDeepCSVJetTags", "probb"), | ||
| edm::InputTag("pfDeepCSVJetTags", "probbb"), | ||
| }); | ||
| temp2.addParameter<std::string>("name", "BvsAll"); | ||
| temp1.push_back(temp2); | ||
| } | ||
| { | ||
| edm::ParameterSet temp2; | ||
| temp2.addParameter<std::vector<edm::InputTag>>( | ||
| "denominator", | ||
| { | ||
| edm::InputTag("pfDeepCSVJetTags", "probc"), | ||
| edm::InputTag("pfDeepCSVJetTags", "probb"), | ||
| edm::InputTag("pfDeepCSVJetTags", "probbb"), | ||
| }); | ||
| temp2.addParameter<std::vector<edm::InputTag>>( | ||
| "numerator", | ||
| { | ||
| edm::InputTag("pfDeepCSVJetTags", "probc"), | ||
| }); | ||
| temp2.addParameter<std::string>("name", "CvsB"); | ||
| temp1.push_back(temp2); | ||
| } | ||
| { | ||
| edm::ParameterSet temp2; | ||
| temp2.addParameter<std::vector<edm::InputTag>>( | ||
| "denominator", | ||
| { | ||
| edm::InputTag("pfDeepCSVJetTags", "probudsg"), | ||
| edm::InputTag("pfDeepCSVJetTags", "probc"), | ||
| }); | ||
| temp2.addParameter<std::vector<edm::InputTag>>( | ||
| "numerator", | ||
| { | ||
| edm::InputTag("pfDeepCSVJetTags", "probc"), | ||
| }); | ||
| temp2.addParameter<std::string>("name", "CvsL"); | ||
| temp1.push_back(temp2); | ||
| } | ||
| desc.addVPSet("discriminators", vpsd1, temp1); | ||
| } | ||
| descriptions.addDefault(desc); | ||
| } | ||
|
|
||
| // define this as a plug-in | ||
| DEFINE_FWK_MODULE(BTagProbabilityToDiscriminator); |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Oops, something went wrong.