thundersvm 0.1.0

checksums.yaml

@@ -0,0 +1,7 @@
+---
+SHA256:
+  metadata.gz: 9f3ccc0ace653a21c742dd36bdf2cf33d941fd7c157b98ea5e453524677dc3da
+  data.tar.gz: 6a35aad608e187dac40f4cb2cc798d8fe6b17bb7617e570861f5a044bfd30f5b
+SHA512:
+  metadata.gz: ca1f18fce237bb032f64e33c36636ff379fb6a267dabcb83f30700a9bfc60c3fd8a1511afbf17cfe5c009bb1c584d2fd463dbba6a35bad7f9a4d5bc9f8b6cbeb
+  data.tar.gz: 775d599e2e0334fe4ae24aa6ae7789087f0c16cc8e9a786fb5b344b784a62b6f509db882ce70dfbecdf0d76e4ed562a8b140972f5c536349c899660cebd78a1a

data/CHANGELOG.md

@@ -0,0 +1,3 @@
+## 0.1.0 (2019-11-24)
+
+- First release

data/LICENSE.txt

@@ -0,0 +1,22 @@
+Copyright (c) 2019 Andrew Kane
+
+MIT License
+
+Permission is hereby granted, free of charge, to any person obtaining
+a copy of this software and associated documentation files (the
+"Software"), to deal in the Software without restriction, including
+without limitation the rights to use, copy, modify, merge, publish,
+distribute, sublicense, and/or sell copies of the Software, and to
+permit persons to whom the Software is furnished to do so, subject to
+the following conditions:
+
+The above copyright notice and this permission notice shall be
+included in all copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
+LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

data/README.md

@@ -0,0 +1,141 @@
+# ThunderSVM
+
+[ThunderSVM](https://github.com/Xtra-Computing/thundersvm) - high-performance parallel SVMs - for Ruby
+
+:fire: Uses GPUs and multi-core CPUs for blazing performance
+
+For a great intro on support vector machines, check out [this video](https://www.youtube.com/watch?v=efR1C6CvhmE).
+
+## Installation
+
+First, [install ThunderSVM](https://github.com/Xtra-Computing/thundersvm/blob/master/docs/how-to.md#install-thundersvm). Add this line to your application’s Gemfile:
+
+```ruby
+gem 'thundersvm'
+```
+
+## Getting Started
+
+Prep your data
+
+```ruby
+x = [[1, 2], [3, 4], [5, 6], [7, 8]]
+y = [1, 2, 3, 4]
+```
+
+Train a model
+
+```ruby
+model = ThunderSVM::Regressor.new
+model.fit(x, y)
+```
+
+Use `ThunderSVM::Classifier` for classification and `ThunderSVM::Model` for other models
+
+Make predictions
+
+```ruby
+model.predict(x)
+```
+
+Save the model to a file
+
+```ruby
+model.save_model("model.txt")
+```
+
+Load the model from a file
+
+```ruby
+model = ThunderSVM.load_model("model.txt")
+```
+
+Get support vectors
+
+```ruby
+model.support_vectors
+```
+
+## Cross-Validation
+
+Perform cross-validation
+
+```ruby
+model.cv(x, y)
+```
+
+Specify the number of folds
+
+```ruby
+model.cv(x, y, folds: 5)
+```
+
+## Parameters
+
+Defaults shown below
+
+```ruby
+ThunderSVM::Model.new(
+  svm_type: :c_svc,    # set type of SVM (c_svc, nu_svc, one_class, epsilon_svr, nu_svr)
+  kernel: :rbf,        # set type of kernel function (linear, polynomial, rbf, sigmoid)
+  degree: 3,           # set degree in kernel function
+  gamma: nil,          # set gamma in kernel function
+  coef0: 0,            # set coef0 in kernel function
+  c: 1,                # set the parameter C of C-SVC, epsilon-SVR, and nu-SVR
+  nu: 0.5,             # set the parameter nu of nu-SVC, one-class SVM, and nu-SVR
+  epsilon: 0.1,        # set the epsilon in loss function of epsilon-SVR
+  max_memory: 8192,    # constrain the maximum memory size (MB) that thundersvm uses
+  tolerance: 0.001,    # set tolerance of termination criterion
+  probability: false,  # whether to train a SVC or SVR model for probability estimates
+  gpu: 0,              # specify which gpu to use
+  cores: nil,          # set the number of cpu cores to use (defaults to all)
+  verbose: false       # verbose mode
+)
+```
+
+## Data
+
+Data can be a Ruby array
+
+```ruby
+[[1, 2], [3, 4], [5, 6], [7, 8]]
+```
+
+Or a Numo array
+
+```ruby
+Numo::DFloat.cast([[1, 2], [3, 4], [5, 6], [7, 8]])
+```
+
+Or the path a file in `libsvm` format (better for sparse data)
+
+```ruby
+model.fit("train.txt")
+model.predict("test.txt")
+```
+
+## Resources
+
+- [ThunderSVM: A Fast SVM Library on GPUs and CPUs](https://github.com/Xtra-Computing/thundersvm/blob/master/thundersvm-full.pdf)
+
+## History
+
+View the [changelog](https://github.com/ankane/thundersvm/blob/master/CHANGELOG.md)
+
+## Contributing
+
+Everyone is encouraged to help improve this project. Here are a few ways you can help:
+
+- [Report bugs](https://github.com/ankane/thundersvm/issues)
+- Fix bugs and [submit pull requests](https://github.com/ankane/thundersvm/pulls)
+- Write, clarify, or fix documentation
+- Suggest or add new features
+
+To get started with development:
+
+```sh
+git clone https://github.com/ankane/thundersvm.git
+cd thundersvm
+bundle install
+bundle exec rake test
+```

data/lib/thundersvm.rb

@@ -0,0 +1,28 @@
+# stdlib
+require "fiddle/import"
+require "fileutils"
+require "tempfile"
+
+# modules
+require "thundersvm/model"
+require "thundersvm/classifier"
+require "thundersvm/regressor"
+require "thundersvm/version"
+
+module ThunderSVM
+  class Error < StandardError; end
+
+  class << self
+    attr_accessor :ffi_lib
+  end
+  self.ffi_lib = ["libthundersvm.so", "libthundersvm.dylib", "thundersvm.dll"]
+
+  # friendlier error message
+  autoload :FFI, "thundersvm/ffi"
+
+  def self.load_model(path)
+    model = Model.new
+    model.load_model(path)
+    model
+  end
+end

data/lib/thundersvm/classifier.rb

@@ -0,0 +1,7 @@
+module ThunderSVM
+  class Classifier < Model
+    def initialize(svm_type: :c_svc, **options)
+      super(svm_type: svm_type, **options)
+    end
+  end
+end

data/lib/thundersvm/ffi.rb

@@ -0,0 +1,19 @@
+module ThunderSVM
+  module FFI
+    extend Fiddle::Importer
+
+    libs = ThunderSVM.ffi_lib.dup
+    begin
+      dlload libs.shift
+    rescue Fiddle::DLError => e
+      retry if libs.any?
+      raise e if ENV["THUNDERSVM_DEBUG"]
+      raise LoadError, "Could not find ThunderSVM"
+    end
+
+    extern "void thundersvm_train(int argc, char **argv)"
+    extern "void thundersvm_train_after_parse(char **option, int len, char *file_name)"
+    extern "void thundersvm_predict(int argc, char **argv)"
+    extern "void thundersvm_predict_after_parse(char *model_file_name, char *output_file_name, char **option, int len)"
+  end
+end

data/lib/thundersvm/model.rb

@@ -0,0 +1,191 @@
+module ThunderSVM
+  class Model
+    def initialize(svm_type: :c_svc, kernel: :rbf, degree: 3, gamma: nil, coef0: 0,
+      c: 1, nu: 0.5, epsilon: 0.1, max_memory: 8192, tolerance: 0.001,
+      probability: false, gpu: 0, cores: nil, verbose: nil)
+
+      @svm_type = svm_type.to_sym
+      @kernel = kernel.to_sym
+      @degree = degree
+      @gamma = gamma
+      @coef0 = coef0
+      @c = c
+      @nu = nu
+      @epsilon = epsilon
+      @max_memory = max_memory
+      @tolerance = tolerance
+      @probability = probability
+      @gpu = gpu
+      @cores = cores
+      @verbose = verbose
+    end
+
+    def fit(x, y = nil)
+      train(x, y)
+    end
+
+    def cv(x, y = nil, folds: 5)
+      train(x, y, folds: folds)
+    end
+
+    def predict(x)
+      dataset_file = create_dataset(x)
+      out_file = create_tempfile
+      argv = ["thundersvm-predict", dataset_file.path, @model_file.path, out_file.path]
+      FFI.thundersvm_predict(argv.size, str_ptr(argv))
+      func = [:c_svc, :nu_svc].include?(@svm_type) ? :to_i : :to_f
+      out_file.each_line.map(&func)
+    end
+
+    def save_model(path)
+      raise Error, "Not trained" unless @model_file
+      FileUtils.cp(@model_file.path, path)
+      nil
+    end
+
+    def load_model(path)
+      @model_file ||= create_tempfile
+      # TODO ensure tempfile is still cleaned up
+      FileUtils.cp(path, @model_file.path)
+      @svm_type = read_header["svm_type"].to_sym
+      @kernel = read_header["kernel_type"].to_sym
+      nil
+    end
+
+    def support_vectors
+      vectors = []
+      sv = false
+      read_txt do |line|
+        if sv
+          index = line.index("1:")
+          vectors << line[index..-1].split(" ").map { |v| v.split(":").last.to_f }
+        elsif line.start_with?("SV")
+          sv = true
+        end
+      end
+      vectors
+    end
+
+    def dual_coef
+      vectors = []
+      sv = false
+      read_txt do |line|
+        if sv
+          index = line.index("1:")
+          line[0...index].split(" ").map(&:to_f).each_with_index do |v, i|
+            (vectors[i] ||= []) << v
+          end
+        elsif line.start_with?("SV")
+          sv = true
+        end
+      end
+      vectors
+    end
+
+    def self.finalize_file(file)
+      # must use proc instead of stabby lambda
+      proc do
+        file.close
+        file.unlink
+      end
+    end
+
+    private
+
+    def train(x, y = nil, folds: nil)
+      dataset_file = create_dataset(x, y)
+      @model_file ||= create_tempfile
+
+      svm_types = {
+        c_svc: 0,
+        nu_svc: 1,
+        one_class: 2,
+        epsilon_svr: 3,
+        nu_svr: 4
+      }
+      s = svm_types[@svm_type]
+      raise Error, "Unknown SVM type: #{@svm_type}" unless s
+
+      kernels = {
+        linear: 0,
+        polynomial: 1,
+        rbf: 2,
+        sigmoid: 3
+      }
+      t = kernels[@kernel]
+      raise Error, "Unknown kernel: #{@kernel}" unless t
+
+      verbose = @verbose
+      verbose = true if folds && verbose.nil?
+
+      argv = ["thundersvm-train"]
+      argv += ["-s", s]
+      argv += ["-t", t]
+      argv += ["-d", @degree.to_i] if @degree
+      argv += ["-g", @gamma.to_f] if @gamma
+      argv += ["-r", @coef0.to_f] if @coef0
+      argv += ["-c", @c.to_f] if @c
+      argv += ["-n", @nu.to_f] if @nu
+      argv += ["-p", @epsilon.to_f] if @epsilon
+      argv += ["-m", @max_memory.to_i] if @max_memory
+      argv += ["-e", @tolerance.to_f] if @tolerance
+      argv += ["-b", @probability ? 1 : 0] if @probability
+      argv += ["-v", folds.to_i] if folds
+      argv += ["-u", @gpu.to_i] if @gpu
+      argv += ["-o", @cores.to_i] if @cores
+      argv << "-q" unless verbose
+      argv += [dataset_file.path, @model_file.path]
+
+      FFI.thundersvm_train(argv.size, str_ptr(argv))
+      nil
+    end
+
+    def create_dataset(x, y = nil)
+      if x.is_a?(String)
+        raise ArgumentError, "Cannot pass y with file" if y
+        File.open(x)
+      else
+        contents = String.new("")
+        y ||= [0] * x.size
+        x.to_a.zip(y.to_a).each do |xi, yi|
+          contents << "#{yi.to_i} #{xi.map.with_index { |v, i| "#{i + 1}:#{v.to_f}" }.join(" ")}\n"
+        end
+        dataset = create_tempfile
+        dataset.write(contents)
+        dataset.close
+        dataset
+      end
+    end
+
+    def str_ptr(arr)
+      ptr = Fiddle::Pointer.malloc(Fiddle::SIZEOF_VOIDP * arr.size)
+      arr.each_with_index do |v, i|
+        ptr[i * Fiddle::SIZEOF_VOIDP, Fiddle::SIZEOF_VOIDP] = Fiddle::Pointer["#{v}\x00"].ref
+      end
+      ptr
+    end
+
+    def create_tempfile
+      file = Tempfile.new("thundersvm")
+      ObjectSpace.define_finalizer(self, self.class.finalize_file(file))
+      file
+    end
+
+    def read_header
+      model = {}
+      read_txt do |line|
+        break if line.start_with?("SV")
+        k, v = line.split(" ", 2)
+        model[k] = v.strip
+      end
+      model
+    end
+
+    def read_txt
+      @model_file.rewind
+      @model_file.each_line do |line|
+        yield line
+      end
+    end
+  end
+end

data/lib/thundersvm/regressor.rb

@@ -0,0 +1,7 @@
+module ThunderSVM
+  class Regressor < Model
+    def initialize(svm_type: :epsilon_svr, **options)
+      super(svm_type: svm_type, **options)
+    end
+  end
+end

data/lib/thundersvm/version.rb

@@ -0,0 +1,3 @@
+module ThunderSVM
+  VERSION = "0.1.0"
+end

metadata

@@ -0,0 +1,107 @@
+--- !ruby/object:Gem::Specification
+name: thundersvm
+version: !ruby/object:Gem::Version
+  version: 0.1.0
+platform: ruby
+authors:
+- Andrew Kane
+autorequire: 
+bindir: bin
+cert_chain: []
+date: 2019-11-25 00:00:00.000000000 Z
+dependencies:
+- !ruby/object:Gem::Dependency
+  name: bundler
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
+- !ruby/object:Gem::Dependency
+  name: rake
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
+- !ruby/object:Gem::Dependency
+  name: minitest
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '5'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '5'
+- !ruby/object:Gem::Dependency
+  name: numo-narray
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
+description: 
+email: andrew@chartkick.com
+executables: []
+extensions: []
+extra_rdoc_files: []
+files:
+- CHANGELOG.md
+- LICENSE.txt
+- README.md
+- lib/thundersvm.rb
+- lib/thundersvm/classifier.rb
+- lib/thundersvm/ffi.rb
+- lib/thundersvm/model.rb
+- lib/thundersvm/regressor.rb
+- lib/thundersvm/version.rb
+homepage: https://github.com/ankane/thundersvm
+licenses:
+- MIT
+metadata: {}
+post_install_message: 
+rdoc_options: []
+require_paths:
+- lib
+required_ruby_version: !ruby/object:Gem::Requirement
+  requirements:
+  - - ">="
+    - !ruby/object:Gem::Version
+      version: '2.4'
+required_rubygems_version: !ruby/object:Gem::Requirement
+  requirements:
+  - - ">="
+    - !ruby/object:Gem::Version
+      version: '0'
+requirements: []
+rubygems_version: 3.0.6
+signing_key: 
+specification_version: 4
+summary: ThunderSVM - high-performance parallel SVMs - for Ruby
+test_files: []