DEPRECATED: Regularised Empirical Risk Minimisation Framework (SVMs, LogReg, Linear Regression) in Julia
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.

README.md

RegERMs Logo

RegERMs.jl

Build Status Coverage Status RegERMs RegERMs

This package implements several machine learning algorithms in a regularised empirical risk minimisation framework (SVMs, LogReg, Linear Regression) in Julia.

Quick start

Some examples:

using RegERMs

# define some toy data (XOR - example)
np = 100
nn = 100
X = [randn(int(np/2),1)+1 randn(int(np/2),1)+1; randn(int(np/2-0.5),1)-1 randn(int(np/2-0.5),1)-1;
     randn(int(nn/2),1)+1 randn(int(nn/2),1)-1; randn(int(nn/2-0.5),1)-1 randn(int(nn/2-0.5),1)+1] # examples with 2 features
y = int(vec([ones(np,1); -ones(nn,1)]))       # binary class values

# use rbf kernel by using mercer map
map = MercerMap(X, :rbf)
X = RegERMs.apply(map)

# choose (linear) SVM as learning algorithm with regularization parameter 0.1
svm = SVM(X, y; λ=0.1)

# get a solution 
model = optimize(svm)

# make predictions and compute accuracy
ybar = predict(model, X)
acc = mean(ybar .== y)

Documentation

Full documentation available at Read the Docs.