ITM Web Conf.
Volume 12, 2017The 4th Annual International Conference on Information Technology and Applications (ITA 2017)
|Number of page(s)||7|
|Section||Session 1: Robotics|
|Published online||05 September 2017|
Accelerating Relevance Vector Machine for Large-Scale Data on Spark
1 National Engineering Laboratory for Fiber Optic Sensing Technology,Wuhan University of Technology, Wuhan 430070, China
2 School of Computer Science and Technology,Wuhan University of Technology, Wuhan 430070, China
Relevance vector machine (RVM) is a machine learning algorithm based on a sparse Bayesian framework, which performs well when running classification and regression tasks on small-scale datasets. However, RVM also has certain drawbacks which restricts its practical applications such as (1) slow training process, (2) poor performance on training large-scale datasets. In order to solve these problem, we propose Discrete AdaBoost RVM (DAB-RVM) which incorporate ensemble learning in RVM at first. This method performs well with large-scale low-dimensional datasets. However, as the number of features increases, the training time of DAB-RVM increases as well. To avoid this phenomenon, we utilize the sufficient training samples of large-scale datasets and propose all features boosting RVM (AFB-RVM), which modifies the way of obtaining weak classifiers. In our experiments we study the differences between various boosting techniques with RVM, demonstrating the performance of the proposed approaches on Spark. As a result of this paper, two proposed approaches on Spark for different types of large-scale datasets are available.
© The Authors, published by EDP Sciences, 2017
This is an Open Access article distributed under the terms of the Creative Commons Attribution License 4.0, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.