rumale-metric_learning 0.25.0 → 0.27.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 8fbc84de9f21a9fda63e31898cf1fa28b2a29dad7f41821f9cf68d64a881906d
-  data.tar.gz: bf66f89e3d23653947e8bcdbd3c5054c88ecaca13b51d26c89cf3764f2fa6179
+  metadata.gz: e7a10afa083e39f86746029f24ca6edd4edc417a045c8608d0c597b788197cf6
+  data.tar.gz: b3b757f6598fbb21d5f6b1383d1bdfa676962ac12136a02c95d5c7dbdb695065
 SHA512:
-  metadata.gz: 4d63cd938c957f3963d4497b7067fc77171f977a079d4ff94a2d82acf0c4a1123d8ac50833437c2c1bfd66d3decaeabaa772e2bf8630dd7ed7443788249d8850
-  data.tar.gz: 97e1cbee0e977bf161fcf82b161e961877afc795eaf2e61475b3d88c2a2f524d3ad4233a9be421d310659d146f94c738ae83c0b4d154f570688671499f22902a
+  metadata.gz: a534929e8338c1513872c432576d2bf4a73fe9ff1999cfa997289626e479ab823dcaeff8217f0aeebee0392f64cc6f9ff1790f52c7047c994d55a7d01b55ae16
+  data.tar.gz: a346e8296fd2c1d28ab014d4869b1fb2876dae2b0874729a1768f13896c531c309fab05f8f0a0780d22f5b738e5b1494736cfb126b81ac395638663e73f78944

data/lib/rumale/metric_learning/local_fisher_discriminant_analysis.rb

@@ -0,0 +1,115 @@
+# frozen_string_literal: true
+
+require 'rumale/base/estimator'
+require 'rumale/base/transformer'
+require 'rumale/validation'
+
+module Rumale
+  module MetricLearning
+    # LocalFisherDiscriminantAnalysis is a class that implements Local Fisher Discriminant Analysis.
+    #
+    # @example
+    #   require 'rumale/metric_learning/local_fisher_discriminant_analysis'
+    #
+    #   transformer = Rumale::MetricLearning::LocalFisherDiscriminantAnalysis.new
+    #   transformer.fit(training_samples, traininig_labels)
+    #   low_samples = transformer.transform(testing_samples)
+    #
+    # *Reference*
+    # - Sugiyama, M., "Local Fisher Discriminant Analysis for Supervised Dimensionality Reduction," Proc. ICML'06, pp. 905--912, 2006.
+    class LocalFisherDiscriminantAnalysis < ::Rumale::Base::Estimator
+      include ::Rumale::Base::Transformer
+
+      # Returns the transform matrix.
+      # @return [Numo::DFloat] (shape: [n_components, n_features])
+      attr_reader :components
+
+      # Return the class labels.
+      # @return [Numo::Int32] (shape: [n_classes])
+      attr_reader :classes
+
+      # Create a new transformer with LocalFisherDiscriminantAnalysis.
+      #
+      # @param n_components [Integer] The number of components.
+      # @param gamma [Float] The parameter of rbf kernel, if nil it is 1 / n_features.
+      def initialize(n_components: nil, gamma: nil)
+        super()
+        @params = {
+          n_components: n_components,
+          gamma: gamma
+        }
+      end
+
+      # Fit the model with given training data.
+      #
+      # @param x [Numo::DFloat] (shape: [n_samples, n_features]) The training data to be used for fitting the model.
+      # @param y [Numo::Int32] (shape: [n_samples]) The labels to be used for fitting the model.
+      # @return [LocalFisherDiscriminantAnalysis] The learned classifier itself.
+      def fit(x, y)
+        unless enable_linalg?(warning: false)
+          raise 'LocalFisherDiscriminatAnalysis#fit requires Numo::Linalg but that is not loaded.'
+        end
+
+        x = Rumale::Validation.check_convert_sample_array(x)
+        y = Rumale::Validation.check_convert_label_array(y)
+        Rumale::Validation.check_sample_size(x, y)
+
+        # initialize some variables.
+        n_samples, n_features = x.shape
+        @classes = Numo::Int32[*y.to_a.uniq.sort]
+        n_components = @params[:n_components] || n_features
+        @params[:gamma] ||= 1.fdiv(n_features)
+        affinity_mat = Rumale::PairwiseMetric.rbf_kernel(x, nil, @params[:gamma])
+        affinity_mat[affinity_mat.diag_indices] = 1.0
+
+        # calculate within and mixture scatter matricies.
+        class_mat = Numo::DFloat.zeros(n_samples, n_samples)
+        within_weight_mat = Numo::DFloat.zeros(n_samples, n_samples)
+        @classes.each do |label|
+          pos = y.eq(label)
+          n_class_samples = pos.count
+          pos_vec = Numo::DFloat.cast(pos)
+          pos_mat = pos_vec.outer(pos_vec)
+          class_mat += pos_mat
+          within_weight_mat += pos_mat * 1.fdiv(n_class_samples)
+        end
+
+        mixture_weight_mat = ((affinity_mat - 1) / n_samples) * class_mat + 1.fdiv(n_samples)
+        within_weight_mat *= affinity_mat
+        mixture_weight_mat = mixture_weight_mat.sum(axis: 1).diag - mixture_weight_mat
+        within_weight_mat = within_weight_mat.sum(axis: 1).diag - within_weight_mat
+
+        # calculate components.
+        mixture_mat = x.transpose.dot(mixture_weight_mat.dot(x))
+        within_mat = x.transpose.dot(within_weight_mat.dot(x))
+        _, evecs = Numo::Linalg.eigh(mixture_mat, within_mat, vals_range: (n_features - n_components)...n_features)
+        comps = evecs.reverse(1).transpose.dup
+        @components = n_components == 1 ? comps[0, true].dup : comps.dup
+        self
+      end
+
+      # Fit the model with training data, and then transform them with the learned model.
+      #
+      # @param x [Numo::DFloat] (shape: [n_samples, n_features]) The training data to be used for fitting the model.
+      # @param y [Numo::Int32] (shape: [n_samples]) The labels to be used for fitting the model.
+      # @return [Numo::DFloat] (shape: [n_samples, n_components]) The transformed data
+      def fit_transform(x, y)
+        x = Rumale::Validation.check_convert_sample_array(x)
+        y = Rumale::Validation.check_convert_label_array(y)
+        Rumale::Validation.check_sample_size(x, y)
+
+        fit(x, y).transform(x)
+      end
+
+      # Transform the given data with the learned model.
+      #
+      # @param x [Numo::DFloat] (shape: [n_samples, n_features]) The data to be transformed with the learned model.
+      # @return [Numo::DFloat] (shape: [n_samples, n_components]) The transformed data.
+      def transform(x)
+        x = Rumale::Validation.check_convert_sample_array(x)
+
+        x.dot(@components.transpose)
+      end
+    end
+  end
+end

data/lib/rumale/metric_learning/version.rb

@@ -5,6 +5,6 @@ module Rumale
   # Module for metric learning algorithms.
   module MetricLearning
     # @!visibility private
-    VERSION = '0.25.0'
+    VERSION = '0.27.0'
   end
 end

data/lib/rumale/metric_learning.rb

@@ -3,6 +3,7 @@
 require 'numo/narray'
 
 require_relative 'metric_learning/fisher_discriminant_analysis'
+require_relative 'metric_learning/local_fisher_discriminant_analysis'
 require_relative 'metric_learning/mlkr'
 require_relative 'metric_learning/neighbourhood_component_analysis'
 require_relative 'metric_learning/version'

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: rumale-metric_learning
 version: !ruby/object:Gem::Version
-  version: 0.25.0
+  version: 0.27.0
 platform: ruby
 authors:
 - yoshoku
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2023-01-18 00:00:00.000000000 Z
+date: 2023-08-26 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: lbfgsb
@@ -44,28 +44,28 @@ dependencies:
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 0.25.0
+        version: 0.27.0
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 0.25.0
+        version: 0.27.0
 - !ruby/object:Gem::Dependency
   name: rumale-decomposition
   requirement: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 0.25.0
+        version: 0.27.0
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 0.25.0
+        version: 0.27.0
 description: |
   Rumale::MetricLearning provides metric learning algorithms,
   such as Fisher Discriminant Analysis and Neighboourhood Component Analysis
@@ -80,6 +80,7 @@ files:
 - README.md
 - lib/rumale/metric_learning.rb
 - lib/rumale/metric_learning/fisher_discriminant_analysis.rb
+- lib/rumale/metric_learning/local_fisher_discriminant_analysis.rb
 - lib/rumale/metric_learning/mlkr.rb
 - lib/rumale/metric_learning/neighbourhood_component_analysis.rb
 - lib/rumale/metric_learning/version.rb