GitHub

eSPA

Documentation for eSPA.

eSPA.eSPA — Module
eSPA.jl

Implementations of eSPA, eSPA+ and GOAl.

Exports:

  • eSPAdiscrete
  • eSPAfuzzy
  • eSPAplus
  • GOAL
  • eSPAhybrid
  • fit!
  • predict
source
eSPA.GOAL — Type
GOAL(K::Int, eps_CL::Float64, G::Int, tol::Float64, max_iter::Int)

Implementation of GOAL [Vecchi+2024]

Fields

  • K::Int: Number of Clusters
  • eps_CL::Float64: Hyperparameter for the classifyer loss
  • G::Int: Dimension of Gauge
  • tol::Float64: Break-condition for optimizaiton
  • max_iter::Int: Maximum number of optimization iterations
source
eSPA.eSPAdiscrete — Type
eSPAdiscrete(K::Int, eps_CL::Float64, eps_E::Float64, tol::Float64, max_iter::Int)

Discrete version of eSPA [Horenko2020]

Fields

  • K::Int: Number of Clusters
  • eps_CL::Float64: Hyperparameter for the classifyer loss
  • eps_E::Float64: Hyperparameter for the feature selection loss
  • tol::Float64: Break-condition for optimizaiton
  • max_iter::Int: Maximum number of optimization iterations
source
eSPA.eSPAfuzzy — Type
eSPAfuzzy(K::Int, eps_CL::Float64, eps_E::Float64, tol::Float64, max_iter::Int)

Fuzzy version of eSPA [Horenko2020]

Fields

  • K::Int: Number of Clusters
  • eps_CL::Float64: Hyperparameter for the classifyer loss
  • eps_E::Float64: Hyperparameter for the feature selection loss
  • tol::Float64: Break-condition for optimizaiton
  • max_iter::Int: Maximum number of optimization iterations
source
eSPA.eSPAhybrid — Type
eSPAhybrid(K::Int, eps_CL::Float64, eps_E::Float64, tol::Float64, max_iter::Int, num_fuzzy_steps::Int)

Fields

  • K::Int: Number of Clusters
  • eps_CL::Float64: Hyperparameter for the classifyer loss
  • eps_E::Float64: Hyperparameter for the feature selection loss
  • tol::Float64: Break-condition for discrete optimizaiton
  • max_iter::Int:
  • num_fuzzy_steps::Int: NUmber of fuzzy steps after discrete steps are finished
source
eSPA.eSPAplus — Type
eSPAplus(K::Int, eps_CL::Float64, eps_E::Float64, tol::Float64, max_iter::Int)

Implementation of eSPA+ [Vecchi+2022]

Fields

  • K::Int: Number of Clusters
  • eps_CL::Float64: Hyperparameter for the classifyer loss
  • eps_E::Float64: Hyperparameter for the feature selection loss
  • tol::Float64: Break-condition for optimizaiton
  • max_iter::Int: Maximum number of optimization iterations
source
eSPA.fit! — Method
fit!(model::GOAL, X::AbstractMatrix, y::AbstractVector)

Train GOAL with Data.

Arguments:

  • model::GOAL: Model instance to train.
  • X::AbstractMatrix: Data matrix which includes the features. Rows contain the features, Columns the data point.
  • y::AbstractVector: Data labels. The labels should be Integers between 1 and M.
source
eSPA.fit! — Method
fit!(model::eSPAdiscrete, X::AbstractMatrix, y::AbstractVector)

Train eSPAdiscrete with Data.

Arguments:

  • model::eSPAdiscrete: Model instance to train.
  • X::AbstractMatrix: Data matrix which includes the features. Rows contain the features, Columns the data point.
  • y::AbstractVector: Data labels. The labels should be Integers between 1 and M.
source
eSPA.fit! — Method
fit!(model::eSPAfuzzy, X::AbstractMatrix, y::AbstractVector)

Train eSPAfuzzy with Data.

Arguments:

  • model::eSPAfuzzy: Model instance to train.
  • X::AbstractMatrix: Data matrix which includes the features. Rows contain the features, Columns the data point.
  • y::AbstractVector: Data labels. The labels should be Integers between 1 and M.
source
eSPA.fit! — Method
fit!(model::eSPAhybrid, X::AbstractMatrix, y::AbstractVector)

Train eSPAhybrid with Data.

Arguments:

  • model::eSPAhybrid: Model instance to train.
  • X::AbstractMatrix: Data matrix which includes the features. Rows contain the features, Columns the data point.
  • y::AbstractVector: Data labels. The labels should be Integers between 1 and M.
source
eSPA.fit! — Method
fit!(model::eSPAplus, X::AbstractMatrix, y::AbstractVector)

Train eSPAplus with Data.

Arguments:

  • model::eSPAplus: Model instance to train.
  • X::AbstractMatrix: Data matrix which includes the features. Rows contain the features, Columns the data point.
  • y::AbstractVector: Data labels. The labels should be Integers between 1 and M.
source
eSPA.predict — Method
predict(model::GOAL, X::AbstractMatrix)

Calculate predictions.

Arguments:

  • model::GOAL: Trained instance of GOAL.
  • X::AbstractMatrix: Data matrix which includes the features. Rows contain the features, Columns the data point.
source
eSPA.predict — Method
predict(model::eSPAdiscrete, X::AbstractMatrix)

Calculate predictions.

Arguments:

  • model::eSPAdiscrete: Trained instance of eSPAdiscrete.
  • X::AbstractMatrix: Data matrix which includes the features. Rows contain the features, Columns the data point.
source
eSPA.predict — Method
predict(model::eSPAfuzzy, X::AbstractMatrix)

Calculate predictions.

Arguments:

  • model::eSPAfuzzy: Trained instance of eSPAfuzzy.
  • X::AbstractMatrix: Data matrix which includes the features. Rows contain the features, Columns the data point.
source
eSPA.predict — Method
predict(model::eSPAhybrid, X::AbstractMatrix)

Calculate predictions.

Arguments:

  • model::eSPAhybrid: Trained instance of eSPAhybrid.
  • X::AbstractMatrix: Data matrix which includes the features. Rows contain the features, Columns the data point.
source
eSPA.predict — Method
predict(model::eSPAplus, X::AbstractMatrix)

Calculate predictions.

Arguments:

  • model::eSPAplus: Trained instance of eSPAplus.
  • X::AbstractMatrix: Data matrix which includes the features. Rows contain the features, Columns the data point.
source