Perceptrons.jl

A package with several types of Perceptron classifiers. Perceptrons are fast classifiers and can be used even for big data. Up to now, this package contains a linear perceptron, voted perceptron and a Kernel perceptron for binary classification problems. This project will have the following perceptron classifiers: Multiclass, Kernel, Structured, Voted, Average and Sparse. Some state-of-the-art must be included after these.

Install

Pkg.add("Perceptrons")

Using

using Perceptrons

Examples

using Perceptrons

# training a linear perceptron (solving the OR problem)
X_train        = [1.0 1.0; 0.0 1.0; 1.0 0.0; 0.0 0.0]
Y_train        = [1; 1; 1; 0.0]
X_test         = [.8 .9; .01 1; .9 0.2; 0.1 0.2]

model          = Perceptrons.fit(X_train,Y_train)
Y_pred         = Perceptrons.predict(model,X_test)

println("[Perceptron] accuracy : $(acc(Y_train,Y_pred))")

# training a voted perceptron (solving the OR problem)
model   = Perceptrons.fit(X_train,Y_train,centralize=true,mode="voted")
Y_pred  = Perceptrons.predict(model,X_test)

println("[Voted Perceptron] accuracy : $(acc(Y_train,Y_pred))")

# training a averaged perceptron (solving the OR problem)
model   = Perceptrons.fit(X_train,Y_train,centralize=true,mode="averaged")
Y_pred  = Perceptrons.predict(model,X_test)

println("[Averaged Perceptron] accuracy : $(acc(Y_train,Y_pred))")

# training a kernel perceptron (solving the XOR problem)
X_train = [1.0 1.0; 0.0 1.0; 1.0 0.0; 0.0 0.0]
Y_train = [0.0 ; 1.0; 1.0; 0.0]
X_test  = X_train .+ .03 # adding noise

model   = Perceptrons.fit(X_train,Y_train,centralize=true,mode="kernel",kernel="rbf",width=.01)
Y_pred  = Perceptrons.predict(model,X_test)

println("[Kernel Perceptron] accuracy : $(acc(Y_train,Y_pred))")


# if you want to save your model
Perceptrons.save(model,filename=joinpath(homedir(),"perceptron_model.jld"))

# if you want to load back your model
model = Perceptrons.load(filename=joinpath(homedir(),"perceptron_model.jld"))

What is Implemented

• Voted Perceptron
• Averaged Perceptron
• Kernel Perceptron
• Linear Perceptron

What is Upcoming

• Multiclass Perceptron
• Structured Perceptron
• Sparse Perceptron

Method Description

• Perceptrons.fit - learns from input data and its related single target

  • X::Matrix{:<AbstractFloat} - A matrix that columns are the features and rows are the samples
  • Y::Vector{:<AbstractFloat} - A vector with float values.
  • copydata::Bool = true: If you want to use the same input matrix or a copy.
  • centralize::Bool = true: If you want to z-score columns. Recommended if not z-scored yet.
  • mode::AbstractString = "linear": modes are "linear", "kernel", "voted" and "averaged" perceptron.
  • kernel::AbstractString = "rbf": If you want to apply a nonlinear Perceptron with gaussian Kernel.
  • width::AbstractFloat = 1.0: Rbf Kernel width (Only if kernel="rbf").
  • alpha::Real = 1.0e-2: learning rate.
  • shuffle_epoch::Bool = true: Shuffle dataset for each epoch. Improves convergency.
  • random_state::Int = 42: Use a seed to force same results trhough the same dataset.
  • max_epochs::Int = 5: Maximum epochs.

• Perceptrons.predict - predicts using the learnt model extracted from fit.

  • model::Perceptrons.Model - A Perceptron model learnt from fit.
  • X::Matrix{:<AbstractFloat} - A matrix that columns are the features and rows are the samples.
  • copydata::Bool = true - If you want to use the same input matrix or a copy.

References

TODO

License

The Perceptrons.jl is free software: you can redistribute it and/or modify it under the terms of the MIT "Expat" License. A copy of this license is provided in LICENSE.md