GMUM.r is a R package with new classification and clustering models. This is being done in a close collaboration between GMUM group members and students.
of your data using both known and novel machine learning methods
using online methods and easily scalable algorithms
letting you choose from a selection of underlying implementation/optimization methods
Cross-entropy clustering (shortly CEC) joins advantages of classical k-means with those of EM. Moreover, contrary to k-means and EM, CEC finds the optimal number of clusters by automatically removing redundant ones.
ability to adapt to clusters of desired shapes
ability to find the optimal number of clusters
Known and widely used Support Vector Machine engines with unique Two Ellipsoids preprocessing.
different SVM libraries
unique data preprocessing
Fast C++ implementation of clustering algorithm Growing Neural Gas.
focused on big datasets
integration with igraph
Feel free to start contributing! We welcome you to cooperate - either in giving us feedback and in sending some pull requests. For more information about tasks, bugs and how to work with us, please check below links.
See how our repository evolved.
Our team. Mouse over a member to see contact details.
dr hab. Igor Podolak
Theory behind the implemented algorithms derives from the work of GMUM researchers. Some of the theoretical basis can be obtained by reading related publications.
GMUM.r is being developed at the Faculty of Mathematics and Computer Science of Jagiellonian University. This project is supervised by researchers from the GMUM group, Group of Machine Learning Research. The group focus on various machine learning models and algorithms, with particular interest in hybrid models, connecting concepts from both classification and clustering methods. For students, this project is an unique occasion to make a first step into true science.