SystemDS: A Declarative Machine Learning System for the End-to-End Data Science Lifecycle

Abstract

Machine learning (ML) applications become increasingly common in many domains. ML systems to execute these workloads include numerical computing frameworks and libraries, ML algorithm libraries, and specialized systems for deep neural networks and distributed ML. These systems focus primarily on efficient model training and scoring. However, the data science process is exploratory, and deals with underspecified objectives and a wide variety of heterogeneous data sources. Therefore, additional tools are employed for data engineering and debugging, which requires boundary crossing, unnecessary manual effort, and lacks optimization across the lifecycle. In this paper, we introduce SystemDS, an open source ML system for the end-to-end data science lifecycle from data integration, cleaning, and preparation, over local, distributed, and federated ML model training, to debugging and serving. To this end, we aim to provide a stack of declarative languages with R-like syntax for the different lifecycle tasks, and users with different expertise. We describe the overall system architecture, explain major design decisions (motivated by lessons learned from Apache SystemML), and discuss key features and research directions. Finally, we provide preliminary results that show the potential of end-to-end lifecycle optimization.

Enabling useful provenance in scripting languages with a human-in-the-loop

“…These tools will often collect provenance information in a degree. Unfortunately, they require programs to be written in a domain-specific-language suitable for their specialized ML systems such as SystemDS [9] and HELIX [45], and they only provide provenance for matrix manipulation operations or high-level conceptual ML operators. They do not have the ability to provide flexible data provenance for general-purpose scripts.…”

Efficient End-to-End AutoML via Scalable Search Space Decomposition

“…In recent years, researchers in the database community have been working on raising the level of abstractions of machine learning (ML) and integrating such functionality into today's data management systems [95,96], e.g., SystemML [25], SystemDS [8], Snorkel [71], ZeroER [91], TFX [5,9], Query 2.0 [92], Krypton [66], Cerebro [67], ModelDB [86], MLFlow [94], Deep-Dive [14], HoloClean [72], EaseML [1], ActiveClean [48], and NorthStar [47]. End-to-end AutoML systems [93,97,33] have been an emerging type of systems that has significantly raised the level of abstractions of building ML applications.…”

VolcanoML: Speeding up End-to-End AutoML via Scalable Search Space Decomposition

“…In recent years, researchers in the database community have been working on raising the level of abstractions of machine learning (ML) and integrating such functionality into today's data management systems, e.g., SystemML [18], SystemDS [5],…”