Workflow provenance in the lifecycle of scientific machine learning. (22nd August 2021)

• Record Type: Journal Article
• Authors: Souza, Renan; Azevedo, Leonardo G.; Lourenço, Vítor; Soares, Elton; Thiago, Raphael; Brandão, Rafael; Civitarese, Daniel; Vital Brazil, Emilio; Moreno, Marcio; Valduriez, Patrick; Mattoso, Marta; Cerqueira, Renato; Netto, Marco A. S.
• Other Names: Wu Chase guestEditor.; Yildirim Tulay guestEditor.; Ivanovic Mirjana guestEditor.; Bellatreche Ladjel guestEditor.; Wyrzykowski Roman guestEditor.; Ciorba Florina M. guestEditor.
• Is Part Of: Concurrency and computation. Volume 34:Number 14(2022)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2021-08-22
• Subjects: artificial intelligence -- e‐Science -- explainability -- lineage -- machine learning lifecycle -- provenance -- reproducibility -- scientific machine learning -- scientific workflow -- taxonomy; Parallel processing (Electronic computers) -- Periodicals; Parallel computers -- Periodicals; 004.35
• DOI: 10.1002/cpe.6544
• Languages: English
• ISSNs: 1532-0626

Abstract

Abstract: Machine learning (ML) has already fundamentally changed several businesses. More recently, it has also been profoundly impacting the computational science and engineering domains, like geoscience, climate science, and health science. In these domains, users need to perform comprehensive data analyses combining scientific data and ML models to provide for critical requirements, such as reproducibility, model explainability, and experiment data understanding. However, scientific ML is multidisciplinary, heterogeneous, and affected by the physical constraints of the domain, making such analyses even more challenging. In this work, we leverage workflow provenance techniques to build a holistic view to support the lifecycle of scientific ML. We contribute with (i) characterization of the lifecycle and taxonomy for data analyses; (ii) design decisions to build this view, with a W3C PROV compliant data representation and a reference system architecture; and (iii) lessons learned after an evaluation in an Oil & Gas case using an HPC cluster with 393 nodes and 946 GPUs. The experiments show that the decisions enable queries that integrate domain semantics with ML models while keeping low overhead (<1%), high scalability, and an order of magnitude of query acceleration under certain workloads against without our representation.