SOM.jl - Kohonen's self-organising maps for Julia
The package provides training and visualisation functions for Kohonen's self-organising maps for Julia. Training functions are implemented in pure Julia, without calling external libraries. Visualisation is implemented by using Python's MatPlotLib.
Self-organising maps
Self-organising maps (also referred to as SOMs or Kohonen maps) are artificial neural networks introduced by Teuvo Kohonen in the 1980s. Despite of their age, SOMs are still widely used as an easy and robust unsupervised learning technique for analysis and visualisation of high-dimensional data.
The SOM algorithm maps high-dimensional vectors into a lower-dimensional grid. Most often the target grid is two-dimensional, resulting into intuitively interpretable maps.
For more details see Kohonen's papers, such as
Teuvo Kohonen, Biological Cybernetics, 43 (1982) p. 59-69 Teuvo Kohonen, Biological Cybernetics, 44 (1982) p. 135-140
Technical details and background can be found in Kohonen's still relevant technical report:
Teuvo Kohonen, Jussi Hynninen, Jari Kangas, and Jorma Laaksonen. SOM_PAK: The Self-Organizing Map Program Package. Technical Report A31, Helsinki University of Technology, Laboratory of Computer and Information Science, FIN-02150 Espoo, Finland, 1996. http://www.cis.hut.fi/research/papers/som_tr96.ps.Z
Installation
Julia> Pkg.add("SOM.jl")
–>
Because of the package is not yet registered, it can be installed from the Julia REPL with:
Pkg.clone("https://github.com/andreasdominik/SOM.jl.git")The package requires DataFrames, Distances, Distributions, ProgressMeter, StatsBase, PyPlot, and PyCall with the Python package matplotlib to be installed. The requirements will be installed automatically by the package manager. Sometimes matplotlib causes problems. The easiest way of manual installation is from the Julia REPL into the default Julia-Python environment via:
ENV["PYTHON"]=""
Pkg.add("Conda")
using Conda
Conda.update()
Conda.add("matplotlib")
Pkg.add("PyCall")
Pkg.build("PyCall")
Pkg.add("PyPlot");