AlfClassifier - Introduce support MlpackClassifier (#9)

Author: hudlijac

include/wif/classifiers/alfClassifier.hpp

@@ -9,7 +9,7 @@
 
 #pragma once
 
-#include "wif/classifiers/scikitMlClassifier.hpp"
+#include "wif/classifiers/genericMlClassifier.hpp"
 #include "wif/filesystem/fileModificationChecker.hpp"
 #include "wif/reporters/unirecReporter.hpp"
 #include "wif/utils/timer.hpp"
@@ -47,7 +47,7 @@ class AlfClassifier : public Classifier {
 	 * @param timerIntervalInSeconds number of seconds between each check of file on disk
 	 */
 	AlfClassifier(
-		ScikitMlClassifier& mlClassifier,
+		GenericMlClassifier& mlClassifier,
 		UnirecReporter& reporter,
 		unsigned timerIntervalInSeconds);
 
@@ -85,7 +85,7 @@ class AlfClassifier : public Classifier {
 
 	bool m_modelReloadNeeded = false;
 	uint64_t m_lastModelLoadTime;
-	ScikitMlClassifier& m_mlClassifier;
+	GenericMlClassifier& m_mlClassifier;
 	UnirecReporter& m_reporter;
 	Timer m_timer;
 };

include/wif/classifiers/genericMlClassifier.hpp

@@ -0,0 +1,40 @@
+/**
+ * @file
+ * @author Jachym Hudlicky <[email protected]>
+ * @brief Generic machine learning classifier interface
+ *
+ * SPDX-License-Identifier: BSD-3-Clause
+ */
+
+#pragma once
+
+#include "wif/classifiers/classifier.hpp"
+
+#include <memory>
+#include <vector>
+
+namespace WIF {
+
+/**
+ * @brief Abstract class specifying interfaces for ML classifiers (ScikitMlClassifier and
+ * MlpackClassifier)
+ *
+ */
+class GenericMlClassifier : public Classifier {
+public:
+	/**
+	 * @brief Return the path of the ML model, which is currently loaded
+	 * @return const std::string& path of the model
+	 */
+	virtual const std::string& getMlModelPath() const noexcept = 0;
+
+	/**
+	 * @brief Reload the model from file, which was set in the constructor
+	 *
+	 * @param logicalName contains the logical name of the trained model. The parameter is used only
+	 * with MlpackClassifier (it is unused with ScikitMlClassifier)
+	 */
+	virtual void reloadModelFromDisk(const std::string& logicalName = "trained_data") = 0;
+};
+
+} // namespace WIF

include/wif/classifiers/mlpackClassifier.hpp

@@ -8,7 +8,7 @@
 
 #pragma once
 
-#include "wif/classifiers/classifier.hpp"
+#include "wif/classifiers/genericMlClassifier.hpp"
 #include "wif/ml/mlpackWrapper.hpp"
 
 #include <memory>
@@ -21,13 +21,13 @@ namespace WIF {
  * @brief Classifier performing ML classification which is interconnected with Mlpack library
  *
  */
-class MlpackClassifier : public Classifier {
+class MlpackClassifier : public GenericMlClassifier {
 public:
 	/**
 	 * @brief Construct a new Mlpack Classifier object
 	 *
-	 * @param path contains the path to the file with the trained model.
-	 * @param logicalName contains the logical name of the trained model.
+	 * @param path contains the path to the file with the trained model
+	 * @param logicalName contains the logical name of the trained model
 	 */
 	MlpackClassifier(const std::string& path, const std::string& logicalName = "trained_data");
 
@@ -44,7 +44,7 @@ class MlpackClassifier : public Classifier {
 	 *
 	 * @param flowFeatures flow features to classify
 	 * @return ClfResult result of the classification, which contains double represention class or
-	 * vector<double> with probabilities for each class (depends on model).
+	 * vector<double> with probabilities for each class (depends on model)
 	 */
 	ClfResult classify(const FlowFeatures& flowFeatures) override;
 
@@ -59,16 +59,16 @@ class MlpackClassifier : public Classifier {
 
 	/**
 	 * @brief Return the path of the ML model, which is currently loaded
-	 * @return const std::string& path of the model.
+	 * @return const std::string& path of the model
 	 */
-	const std::string getMlModelPath() const noexcept;
+	const std::string& getMlModelPath() const noexcept override;
 
 	/**
 	 * @brief Reload the model from file, which was set in the constructor
 	 *
-	 * @param logicalName contains the logical name of the trained model.
+	 * @param logicalName contains the logical name of the trained model
 	 */
-	void reloadModelFromDisk(const std::string& logicalName = "trained_data");
+	void reloadModelFromDisk(const std::string& logicalName = "trained_data") override;
 
 private:
 	/**

include/wif/classifiers/scikitMlClassifier.hpp

@@ -8,7 +8,7 @@
 
 #pragma once
 
-#include "wif/classifiers/classifier.hpp"
+#include "wif/classifiers/genericMlClassifier.hpp"
 #include "wif/ml/scikitMlWrapper.hpp"
 
 #include <memory>
@@ -19,7 +19,7 @@ namespace WIF {
 /**
  * @brief Classifier performing Machine-Learning based detection via Scikit-learn library
  */
-class ScikitMlClassifier : public Classifier {
+class ScikitMlClassifier : public GenericMlClassifier {
 public:
 	/**
 	 * @brief Construct a new Scikitlearn Ml Classifier object
@@ -34,7 +34,7 @@ class ScikitMlClassifier : public Classifier {
 	 *
 	 * @return const std::string& the path to the used ML model
 	 */
-	const std::string& getMlModelPath() const noexcept;
+	const std::string& getMlModelPath() const noexcept override;
 
 	/**
 	 * @brief Set feature IDs which will be used for classification
@@ -45,8 +45,10 @@ class ScikitMlClassifier : public Classifier {
 
 	/**
 	 * @brief Reload used ML model from disk
+	 *
+	 * @param logicalName is unused
 	 */
-	void reloadModelFromDisk();
+	void reloadModelFromDisk([[maybe_unused]] const std::string& logicalName) override;
 
 	/**
 	 * @brief Classify single flowFeature object

include/wif/ml/mlpackModels/mlpackModel.hpp

@@ -85,7 +85,7 @@ class MlpackModel {
 	 * @brief Getter for path of the used ML model
 	 * @return const std::string&
 	 */
-	std::string getPath() const;
+	const std::string& getPath() const;
 
 	/**
 	 * @brief Set feature IDs which will be used for classification

include/wif/ml/mlpackWrapper.hpp

@@ -60,7 +60,7 @@ class MlpackWrapper {
 	 * @brief Getter for path of the used ML model
 	 * @return const std::string&
 	 */
-	const std::string getModelPath() const;
+	const std::string& getModelPath() const;
 
 	/**
 	 * @brief Load the model from the file

src/wif/classifiers/alfClassifier.cpp

@@ -25,14 +25,14 @@ void AlfClassifier::AlfCallback::onTick()
 }
 
 AlfClassifier::AlfClassifier(
-	ScikitMlClassifier& mlClassifier,
+	GenericMlClassifier& mlClassifier,
 	UnirecReporter& reporter,
 	unsigned timerIntervalInSeconds)
 	: m_mlClassifier(mlClassifier)
 	, m_reporter(reporter)
 	, m_timer(
 		  timerIntervalInSeconds,
-		  std::make_unique<AlfCallback>(mlClassifier.getMlModelPath(), *this))
+		  std::make_unique<AlfCallback>(m_mlClassifier.getMlModelPath(), *this))
 {
 	updateLastModelLoadTime();
 	m_timer.start();

src/wif/classifiers/mlpackClassifier.cpp

@@ -32,7 +32,7 @@ MlpackClassifier::classify(const std::vector<FlowFeatures>& burstOfFlowFeatures)
 	return m_mlpackWrapper->classify(burstOfFlowFeatures);
 }
 
-const std::string MlpackClassifier::getMlModelPath() const noexcept
+const std::string& MlpackClassifier::getMlModelPath() const noexcept
 {
 	return m_mlpackWrapper->getModelPath();
 }

src/wif/classifiers/scikitMlClassifier.cpp

@@ -28,7 +28,7 @@ void ScikitMlClassifier::setFeatureSourceIDs(const std::vector<FeatureID>& sourc
 	m_scikitMlWrapper->setFeatureSourceIDs(sourceFeatureIDs);
 }
 
-void ScikitMlClassifier::reloadModelFromDisk()
+void ScikitMlClassifier::reloadModelFromDisk([[maybe_unused]] const std::string& logicalName)
 {
 	m_scikitMlWrapper->reloadModel();
 }

src/wif/ml/mlpackModels/mlpackModel.cpp

@@ -15,7 +15,7 @@ bool MlpackModel::isLoaded() const
 	return m_loaded;
 }
 
-std::string MlpackModel::getPath() const
+const std::string& MlpackModel::getPath() const
 {
 	return m_modelPath;
 }