rumale-manifold 0.27.0 → 0.28.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 1f0517a47a43a3554a1f5cb49c69252a14baff20c89b592c0e53e1682edfdb0f
-  data.tar.gz: a01217a27d69b48e7e45b78597f5f3e8aeb8fe3efb147448326a091e29e348e9
+  metadata.gz: 146643ddd999165173dcb3388d01240400288e9093d2d599aeff4348486eb8ce
+  data.tar.gz: bc2c0eca2e8acfa5c07a28dd7b13283ba505c69c066ae94139ac0dc4033933e5
 SHA512:
-  metadata.gz: dfa98f0d332a640f8e5e26ba29393b4fe4084314b22745a07ea0af219239265b301ff6175b175dcc405f242c248f075ae8c106b931d91ff0cd81c938f43b985a
-  data.tar.gz: 844ed666712a82a8769b4479c1794a014d8ed131acf8105d4c9fbc83755b89a638192cf8dd090023f23100cd269d2e6988c5411a9bed8e3f607b286a72308174
+  metadata.gz: 308de32a6c4870c308f8da145a5cd4f6e3558d74457ea0aeb8dbc21c06babbdbb52a97e3dbe6a53b1d01cb622aebad79c69b612737023fbef24ef2a83be422f7
+  data.tar.gz: 6abd9ef1208b3eee3265bc23b4d785e5e70bf34bd88cd05cd1df457dfc02dc7757de63407ee338cb77d818ca2e7d238b50d14b97d9fb76a687511366ebcf7988

data/lib/rumale/manifold/hessian_eigenmaps.rb

@@ -0,0 +1,151 @@
+# frozen_string_literal: true
+
+require 'rumale/base/estimator'
+require 'rumale/base/transformer'
+require 'rumale/pairwise_metric'
+require 'rumale/validation'
+
+module Rumale
+  module Manifold
+    # HessianEigenmaps is a class that implements Hessian Eigenmaps.
+    #
+    # @example
+    #   require 'numo/linalg/autoloader'
+    #   require 'rumale/manifold/hessian_eigenmaps'
+    #
+    #   hem = Rumale::Manifold::HessianEigenmaps.new(n_components: 2, n_neighbors: 15)
+    #   z = hem.fit_transform(x)
+    #
+    # *Reference*
+    # - Donoho, D. L., and Grimes, C., "Hessian eigenmaps: Locally linear embedding techniques for high-dimensional data," Proc. Natl. Acad. Sci. USA, vol. 100, no. 10, pp. 5591--5596, 2003.
+    class HessianEigenmaps < Rumale::Base::Estimator
+      include Rumale::Base::Transformer
+
+      # Return the data in representation space.
+      # @return [Numo::DFloat] (shape: [n_samples, n_components])
+      attr_reader :embedding
+
+      # Create a new transformer with Hessian Eigenmaps.
+      #
+      # @param n_components [Integer] The number of dimensions on representation space.
+      # @param n_neighbors [Integer] The number of nearest neighbors for k-nearest neighbor graph construction.
+      # @param reg_param [Float] The reguralization parameter for local gram matrix in transform method.
+      def initialize(n_neighbors: 5, n_components: 2, reg_param: 1e-6)
+        super()
+        @params = {
+          n_neighbors: n_neighbors,
+          n_components: n_components,
+          reg_param: reg_param
+        }
+      end
+
+      # Fit the model with given training data.
+      #
+      # @overload fit(x) -> LocallyLinearEmbedding
+      #   @param x [Numo::DFloat] (shape: [n_samples, n_features]) The training data to be used for fitting the model.
+      #   @return [LocallyLinearEmbedding] The learned transformer itself.
+      def fit(x, _y = nil) # rubocop:disable Metrics/AbcSize
+        raise 'HessianEigenmaps#fit requires Numo::Linalg but that is not loaded' unless enable_linalg?(warning: false)
+
+        x = Rumale::Validation.check_convert_sample_array(x)
+
+        n_samples = x.shape[0]
+        distance_mat = Rumale::PairwiseMetric.squared_error(x)
+        neighbor_ids = neighbor_ids(distance_mat, @params[:n_neighbors], true)
+
+        tri_n_components = @params[:n_components] * (@params[:n_components] + 1) / 2
+        hessian_mat = Numo::DFloat.zeros(n_samples * tri_n_components, n_samples)
+        ones = Numo::DFloat.ones(@params[:n_neighbors], 1)
+        n_samples.times do |i|
+          tan_coords = tangent_coordinates(x[neighbor_ids[i, true], true])
+          xi = Numo::DFloat.zeros(@params[:n_neighbors], tri_n_components)
+          @params[:n_components].times do |m|
+            offset = Array.new(m + 1) { |v| v }.sum
+            (@params[:n_components] - m).times do |n|
+              xi[true, m * @params[:n_components] - offset + n] = tan_coords[true, m] * tan_coords[true, m + n]
+            end
+          end
+
+          xt, = Numo::Linalg.qr(Numo::DFloat.hstack([ones, tan_coords, xi]))
+          pii = xt[true, (@params[:n_components] + 1)..-1]
+          tri_n_components.times do |j|
+            pj_sum = pii[true, j].sum
+            normalizer = pj_sum <= 1e-8 ? 1 : 1.fdiv(pj_sum)
+            hessian_mat[i * tri_n_components + j, neighbor_ids[i, true]] = pii[true, j] * normalizer
+          end
+        end
+
+        kernel_mat = hessian_mat.transpose.dot(hessian_mat)
+        _, eig_vecs = Numo::Linalg.eigh(kernel_mat, vals_range: 1...(1 + @params[:n_components]))
+
+        @embedding = @params[:n_components] == 1 ? eig_vecs[true, 0].dup : eig_vecs.dup
+        @x_train = x.dup
+
+        self
+      end
+
+      # Fit the model with training data, and then transform them with the learned model.
+      #
+      # @overload fit_transform(x) -> Numo::DFloat
+      #   @param x [Numo::DFloat] (shape: [n_samples, n_features]) The training data to be used for fitting the model.
+      #   @return [Numo::DFloat] (shape: [n_samples, n_components]) The transformed data
+      def fit_transform(x, _y = nil)
+        unless enable_linalg?(warning: false)
+          raise 'HessianEigenmaps#fit_transform requires Numo::Linalg but that is not loaded'
+        end
+
+        fit(x)
+
+        @embedding.dup
+      end
+
+      # Transform the given data with the learned model.
+      # For out-of-sample data embedding, the same method as Locally Linear Embedding is used.
+      #
+      # @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)
+
+        n_samples = x.shape[0]
+        tol = @params[:reg_param].fdiv(@params[:n_neighbors])
+        distance_mat = Rumale::PairwiseMetric.squared_error(x, @x_train)
+        neighbor_ids = neighbor_ids(distance_mat, @params[:n_neighbors], false)
+        weight_mat = Numo::DFloat.zeros(n_samples, @x_train.shape[0])
+
+        n_samples.times do |n|
+          x_local = @x_train[neighbor_ids[n, true], true] - x[n, true]
+          gram_mat = x_local.dot(x_local.transpose)
+          gram_mat += tol * weight_mat.trace * Numo::DFloat.eye(@params[:n_neighbors])
+          weights = Numo::Linalg.solve(gram_mat, Numo::DFloat.ones(@params[:n_neighbors]))
+          weights /= weights.sum + 1e-8
+          weight_mat[n, neighbor_ids[n, true]] = weights
+        end
+
+        weight_mat.dot(@embedding)
+      end
+
+      private
+
+      def neighbor_ids(distance_mat, n_neighbors, contain_self)
+        n_samples = distance_mat.shape[0]
+        neighbor_ids = Numo::Int32.zeros(n_samples, n_neighbors)
+        if contain_self
+          n_samples.times { |n| neighbor_ids[n, true] = (distance_mat[n, true].sort_index.to_a - [n])[0...n_neighbors] }
+        else
+          n_samples.times { |n| neighbor_ids[n, true] = distance_mat[n, true].sort_index.to_a[0...n_neighbors] }
+        end
+        neighbor_ids
+      end
+
+      def tangent_coordinates(x)
+        m = x.mean(axis: 0)
+        cx = x - m
+        cov_mat = cx.transpose.dot(cx)
+        n_features = x.shape[1]
+        _, evecs = Numo::Linalg.eigh(cov_mat, vals_range: (n_features - @params[:n_components])...n_features)
+        cx.dot(evecs.reverse(1))
+      end
+    end
+  end
+end

data/lib/rumale/manifold/version.rb

@@ -5,6 +5,6 @@ module Rumale
   # Module for data embedding algorithms.
   module Manifold
     # @!visibility private
-    VERSION = '0.27.0'
+    VERSION = '0.28.1'
   end
 end

data/lib/rumale/manifold.rb

@@ -4,6 +4,7 @@ require 'numo/narray'
 
 require_relative 'manifold/laplacian_eigenmaps'
 require_relative 'manifold/locally_linear_embedding'
+require_relative 'manifold/hessian_eigenmaps'
 require_relative 'manifold/mds'
 require_relative 'manifold/tsne'
 require_relative 'manifold/version'

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: rumale-manifold
 version: !ruby/object:Gem::Version
-  version: 0.27.0
+  version: 0.28.1
 platform: ruby
 authors:
 - yoshoku
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2023-08-26 00:00:00.000000000 Z
+date: 2023-12-24 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: numo-narray
@@ -30,31 +30,32 @@ dependencies:
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 0.27.0
+        version: 0.28.1
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 0.27.0
+        version: 0.28.1
 - !ruby/object:Gem::Dependency
   name: rumale-decomposition
   requirement: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 0.27.0
+        version: 0.28.1
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 0.27.0
+        version: 0.28.1
 description: |
   Rumale::Manifold provides data embedding algorithms,
-  such as Multi-dimensional Scaling and t-distributed Stochastic Neighbor Embedding,
+  such as Multi-dimensional Scaling, Locally Linear Embedding, Laplacian Eigenmaps, Hessian Eigenmaps,
+  and t-distributed Stochastic Neighbor Embedding,
   with Rumale interface.
 email:
 - yoshoku@outlook.com
@@ -65,6 +66,7 @@ files:
 - LICENSE.txt
 - README.md
 - lib/rumale/manifold.rb
+- lib/rumale/manifold/hessian_eigenmaps.rb
 - lib/rumale/manifold/laplacian_eigenmaps.rb
 - lib/rumale/manifold/locally_linear_embedding.rb
 - lib/rumale/manifold/mds.rb
@@ -93,7 +95,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.3.26
+rubygems_version: 3.4.22
 signing_key:
 specification_version: 4
 summary: Rumale::Manifold provides data embedding algorithms with Rumale interface.