| |
Accelerating Machine Learning Algorithms with GPUs
|
Speaker (s):
Zeyi WEN
Research Fellow,
Department of Computer Science,
School of Computing,
National University of Singapore
|
|
Date:
Time:
Venue:
|
|
October 17, 2018, Wednesday
1:30pm - 2:30pm
Meeting Room 5.1, Level 5
School of Information Systems
Singapore Management University
80 Stamford Road
Singapore 178902
|
|
ABSTRACT
The recent success of machine learning is not only due to more effective algorithms, but also more efficient systems and implementations. Among the machine learning algorithms, Support Vector Machines (SVMs) and decision trees are two widely used algorithms in the data science community according to a survey [1] conducted by Kaggle in 2017. In this seminar, I will first present our open source project ThunderSVM which exploits Graphics Processing Units (GPUs). ThunderSVM supports all the functionalities of LibSVM (including classification, regression and distribution estimation), and is often 10 to 100 times faster than LibSVM. The key techniques and experimental results will be presented. Then, I will talk about our recently developed techniques for Gradient Boosting Decision Tree (GBDT) training on GPUs, and highlight some initial experimental results. This work on GBDTs is part of our ongoing ThunderGBM project which aims to support the major decision tree based algorithms (e.g., GBDTs and random forests). Finally, I will conclude the talk with an ambitious goal of building ThunderML---a workbench for applying machine learning algorithms efficiently and easily.
[1] Kaggle 2017 Survey Results: https://www.kaggle.com/amberthomas/kaggle-2017-survey-results
About the Speaker
Zeyi WEN is currently a research fellow at National University of Singapore. Before working in Singapore, he was a research fellow at The University of Melbourne from 2015 to 2016, and completed his PhD degree at The University of Melbourne in 2015. Zeyi's areas of research include Data Mining, Machine Learning and Parallel Computing. Please visit his homepage for more information: https://www.comp.nus.edu.sg/~wenzy/
|
