Authors: Oliver Albertini, Divya Bhargov, Alexander Denissov, Francisco Guerrero, Nandish Jayaram, Nikhil Kak, Ekta Khanna, Orhan Kislal, Arun Kumar, Frank McQuillan, Lisa Owen, Venkatesh Raghavan, Domino Valdano, Yuhao Zhang
Abstract: Artificial neural networks can be used to create highly accurate models in domains such as language processing and image recog- nition. For example, convolutional neural networks (CNN) can be used to compare satellite images taken over time to monitor changes in rainforest cover in the Amazon basin, or to assess the damage caused by wildfires in order to help direct relief and conser- vation efforts. To address these important problems, we need tools that allow subject matter experts to build models and query image data stored in heterogeneous sources, and the ability to expand or contract computational resources as data volumes change. In this demonstration, we showcase the use of Greenplum as an end-to- end platform for deep learning. We cover the Pivotal Extension Framework (PXF) to fetch and transform images from cloud object stores, and exploit graphics processing unit (GPU) cycles to run Apache MADlib deep learning methods.
Model Hopper Diagram