Go to the list of all software

mySVM/db - Support Vector Machine for Relational Databases


mySVM/db is an implementation of the Support Vector Machine introduced by V. Vapnik (see [Vapnik/98a]). It is a further developement of mySVM and designed to run directly inside a relational database using an internal JAVA engine. The original algorithm is based on the optimization algorithm of SVMlight as described in [Joachims/99a]. mySVM/db provides code for both pattern recognition and regression analysis.

The goal of mySVM/db is to provide users of data-mining applications with an easy to use SVM implementation that makes use of the numerous advantages for handling large amounts of data - like usability, managability, security and standardized interfaces - that modern database systems offer. On the downside, please not that the direct connection of the program to a database server has a serious impact on runtime. So, if performance is your top goal, you should export your data from the database an run some SVM software (e.g. mySVM) on an own machine.

Please note that for now, mySVM/db should be still regarded as beta software. Do not run it on production servers unless you have tested it for your purposes thoroughly. In particular, mySVM/db was designed such that it only needs read access to the input data, so you should run it with minimal access rights. In conclusion, as the license says: The author is not responsible for implications from the use of this software.

Software File:


RĂ¼ping, Stefan


SFB 475 subproject A4