Computer Science Seminar
An Overview of Gradient Descent Optimization Algorithms with
Implementation in MapReduce Programming Model
When: Monday, February 17, 2020
Where: PGH 232
Time: 11:00 AM
Speaker: Dr. Kia Teymourian, Boston University Metropolitan College
Host: Dr. Larry Weidong Shi
Gradient descent optimization algorithms are the most widely used optimization methods in training machine learning models from massive data sets. A large portion of computation time during model training is spent on running the optimizer to find the nearest optimal model parameters. Because of this fact, data scientists and machine learning application developers must have a good level of understanding of these optimization algorithms. The efficient implementation and application of optimization algorithms can highly reduce the required computation time of cluster computers. In this talk, we will provide an introduction to gradient descent algorithms and describe the important algorithmic details. We will learn about variations of these algorithms and discuss the main strengths and issues. We will also walk through some implementation code of gradient descent variations using MapReduce programming model in Apache Spark.Bio:
Dr. Kia Teymourian is an assistant professor in the Department of Computer Science of Boston University's Metropolitan College, where he has been since 2017. From 2014 to 2017 he worked as a postdoctoral research associate at the department of computer science at Rice University. Dr. Teymourian holds a BSc and MSc degree from Berlin University of Technology (TU-Berlin) and a Ph.D. from Freie Universität Berlin. His research interest lies in extended database technologies, data stream processing, big data analytics, as well as web and text mining. He has published more than 40 publications in these fields and made important contributions to multiple large and international research projects, including several funded by the European Commission, the German Federal Ministry of Education and Research (BMBF), and the DARPA Pliny Project at Rice University. At Boston University, Dr. Teymourian teaches courses related to data analysis and Big Data processing, as well as software engineering.