DeepForge: An Open Source, Collaborative Environment for Reproducible Deep Learning

Brian Broll, Jimmy Whitaker

Abstract: In this paper we introduce DeepForge, a platform for deep learning designed to lower the barrier to entry and facilitate the rapid development of deep learning models while maintaining a high degree of transparency and interoperability. Utilizing a cloud-based infrastructure, DeepForge facilitates rapid development by promoting reproducibility, collaboration and remote execution of machine learning pipelines. DeepForge represents an interdisciplinary approach to facilitating deep learning development as it leverages the strengths of Model Integrated Computing to provide a powerful hybrid textual-visual programming platform beneficial to both novices and researchers.

TL;DR: An open source development environment for deep learning designed for simplicity, collaboration and reproducibility of experiments

Keywords: Online Platform, Reproducible Research, Machine Learning, Pattern Recognition