Software

Dear all,

we would like to introduce the new version 2.2 of YALE (Yet Another Learning Environment) to you:

http://yale.cs.uni-dortmund.de

What is Yale?

YALE is a flexible and modular environment for machine learning experiments and data mining applications. The experimental setup can be easily described in a simple XML format or designed with the intuitive graphical user interface (GUI). Yale can be easily extended and integrated into your own application.

Yale includes over 70 different operators for training, validation, feature selection and generation, and preprocessing plus all classifiers, clusterers and association learners available from the completely integrated Weka package.

It reads several data formats including arff, csv, and sparse. Additionally Yale provides direct database access.

What is new?

• New operators: Java versions of MySVM and MyKLR, Meta learner, loaders for attribute sets and attribute weights, NoiseOperator, FourierTransform of features into label space, several new feature selection and generation operators

• Operators can now be disabled which allows more flexible experiment design. Breakpoints are saved now.

• Meta learner schemes can be used to create new attributes from the predictions of learned models

• New system for objects which can be averaged like performance criteria or weights

• Enhanced online 2D and 3D plotting of performance and other results like data or SVM models

Additionally to the Yale core operators two new plugins were released:

• Value Series plugin: provides over 120 mathematical operations on value series like basis transformations or feature extraction. This plugin include operators to handle music data.

• Word Vector Tool plugin: loads text data from several formats and creates word vectors. The possible steps for vector creation include loading (from URL, files,...), inputfilter, charmapper, tokenizer, wordfilter (stop words), stemmer (Lovins, Porter, Snowball), and vector creation (TFIDF, binary,...)

classification and regression tasks, feature selection, generation, and extraction, concept drift, time series analysis, and text mining.

YALE is a freely available open source software under the terms of the GNU General Public License. Since YALE is entirely written in Java, it runs on any major platform/operating system. You are welcome to use it!

Best regards,

Simon Fischer, Ingo Mierswa, Oliver Ritthoff, and Ralf Klinkenberg