Keywords: machine learning, pattern recognition, classification, regression, clustering, density estimation, Python programming
Contact
Address
Universität Erlangen-Nürnberg
Chair of Computer Science 5 (Pattern Recognition)
Martensstr. 3
91058 Erlangen
Germany
Chair of Computer Science 5 (Pattern Recognition)
Martensstr. 3
91058 Erlangen
Germany
Powered by
Dept. of Computer Sc. » Pattern Recognition » Our Team » Steidl, Stefan » Python Classification Toolbox
PD Dr.-Ing. habil. Stefan Steidl
Alumnus of the Pattern Recognition Lab of the Friedrich-Alexander-Universität Erlangen-Nürnberg
Applying mathematically founded machine learning and pattern recognition techniques to fascinating interdisciplinary research areas: from medical image segmentation to computer-assisted diagnosis and therapy control for pathological speech to vocal emotion recognition for human-machine interaction.
Python Classification Toolbox
Study popular machine learning algorithms and create your own implementations in Python for a deeper understanding of the algorithms!
Objectives
Use the Python Classification Toolbox
- as a tool to study existing (and implemented) algorithms for classification, regression, clustering, and density estimation:
- Use the graphical user interface to interactively create and move samples in a 2-dimensional feature space and study the resulting decision boundaries, regression lines, clustering results, and probability density functions.
- Learn how the parameters of the algorithms influence the outcome.
- Download the complete classification toolbox as a binary for Windows or Linux. This version includes a Python environment and all required runtime libraries, but it does not include the Python source code of the toolbox.
- as an object-oriented programming framework to implement your own machine learning algorithms:
- For our pattern recognition courses, we will provide a basic framework including the code for the graphical user interface and the program calls of (some) standard scikit-learn algorithms. However, own implementations are missing as they are part of the class assignments.
- In order to work with this version, you need a running Python 3.x installation and you need to install the Python packages NumPy, SciPy, PyQt4, scikit-learn, and cvxopt.
The Python Classification Toolbox is an open-source alternative to existing similar classification toolboxes for Matlab, focusing on visualization and own implementations of the algorithms. It is intended for 2-dimensional data and rather small data sets only. It is not meant to be a replacement for powerful machine learning toolboxes such as scikit-learn (Python) or WEKA (Java).
Copyright
Copyright 2016 Stefan Steidl
Friedrich-Alexander-Universität Erlangen-Nürnberg
Lehrstuhl für Informatik 5 (Mustererkennung)
Martensstraße 3, 91058 Erlangen, GERMANY
stefan.steidl@fau.de
License
The Python Classification Toolbox is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.
The Python Classification Toolbox is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
See the GNU General Public License for more details.
Downloads
| Version | Platform | Last update |
| Binaries of the Python Classification Toolbox (complete toolbox without source code) | June 13, 2016 | |
| Programming framework (source code, code of class assignments missing) | all platforms | see course Pattern Analysis |
The Windows binary is 32-bit and should work on all Windows versions.
The Linux binary is tested on Ubuntu 14.04 LTS and Debian 8 (jessie). It crashes on openSUSE 13.2.
Use in your own courses:
If you are a lecturer and would like to use this toolbox in your own courses, please send me an e-mail using your official e-mail address of your university. We will be happy to share the complete Python code with you.