Much of modern PCs are equipped with multi-core processors. In fact, the computer operates as if had multiple processors. Some also big servers in particular, have effectively. Software and data mining algorithms must be developed in order to benefit. Currently, few tools are widely available that exploit these new features of machines.
Indeed, the case is not simple. It is impossible to develop a generic approach that would be valid regardless of the learning method used. For a given technology, decompose an algorithm into tasks that can execute in parallel is a research field in itself. Scientific publications full of proposals of all kinds, both methodology (modified algorithm) at the technological level (implementation on machines). A large majority of them are mainly interested in the implementation of large systems. There are very few proposals for lightweight solutions that can be easily introduced on software for personal computers.
In this tutorial, a solution based on threads is highlighted. It is located in version 3.5 of Sipina.
Keywords: multithreading, thread, threads, decision trees, CHAID, SIPINA 3.5, 2.2.2 Knime, RapidMiner 5.0.011
Tutorial: fr_sipina_multithreading.pdf
Data : covtype.arff.zip
References:
Wikipedia, "Decision trees "
Aldinucci, Ruggieri Torquati, "Porting Decision Tree Algorithms to Multicore using FastFlow ", PKDD-2010.
0 comments:
Post a Comment