tzulitai / distributed-svm

Distributed SVM approaches implemented with MapReduce.

Distributed SVM

MapReduce implementations of some distributed SVM training approaches.

Current implemented simple approaches include:

1. The Cascade SVM
2. Bagging-Like SVM

Both approaches refer to this paper:

O. Meyer et al., "Support Vector Machines of Large Data Sets: Simple Parallel Approaches." Data Analysis, Machine Learning, and Knowledge Discovery, Springer, 87-95, 2013

Also, we have used the LIBSVM library for the core training procedures: Reference:

C.-C. Chang and C.-J. Lin. LIBSVM : a library for support vector machines. ACM Transactions on Intelligent Systems and Technology, 2:27:1--27:27, 2011.

Version Info

• Hadoop Version: 2.4.1
• LIBSVM Version: 3.17

Build Info

git clone https://github.com/tzulitai/distributed-svm.git
cd distributed-svm
mvn clean install