We set up Dice as a machine learning and data mining framework and algorithm library for both academia and industry. At the beginning, we focus on building random learning methods, such as Random Decision Trees or RDT (Fan et al, ICDM 2003), and its variations. This is the reason that the project is named "Dice" to denote its stochastic nature. In later stages, we will extend these algorithms from single computing node to distributed computing clusters.
The license of Dice project is Apache 2.0 license. The link of the license is here .
Besides, we hope people who use Dice in papers and applications send feedback to us.
Xiatian Zhang, Tencent.Inc
Wei Fan, IBM T.J Watson Research Center
Click here to download Dice 0.00.0.
Please see the examples in the package dice.examples. These examples demonstrate how to perform binary classification, multi-label classification , and regression problems..
Javadoc of Dice is here. It also can be found in the zip file. Please find them in fold "/doc" of the zip file.
There are 5 data files collected in fold "/data" in the zip file.
Email: dicesproject(AT)gmail.com
Note:dices are NOT dice in the email address.