Call for Papers

The intent of this workshop is to bring together researchers, practitioners, and scientific communities to discuss methods that utilize extreme scale systems for machine learning. This workshop will focus on the greatest challenges in utilizing HPC for machine learning and methods for exploiting data parallelism, model parallelism, ensembles, and parameter search. We invite researchers and practitioners to participate in this workshop to discuss the challenges in using HPC for machine learning and to share the wide range of applications that would benefit from HPC powered machine learning.

In recent years, the models and data available for machine learning (ML) applications have grown dramatically. High performance computing (HPC) offers the opportunity to accelerate performance and deepen understanding of large data sets through machine learning. Current literature and public implementations focus on either cloud-­‐based or small-­‐scale GPU environments. These implementations do not scale well in HPC environments due to inefficient data movement and network communication within the compute cluster, originating from the significant disparity in the level of parallelism. Additionally, applying machine learning to extreme scale scientific data is largely unexplored. To leverage HPC for ML applications, serious advances will be required in both algorithms and their scalable, parallel implementations.

Topics will include but will not be limited to:

• Machine learning models, including deep learning, for extreme scale systems
• Enhancing applicability of machine learning in HPC (e.g. feature engineering, usability)
• Learning large models/optimizing hyper parameters (e.g. deep learning, representation learning)
• Facilitating very large ensembles in extreme scale systems
• Training machine learning models on large datasets and scientific data
• Overcoming the problems inherent to large datasets (e.g. noisy labels, missing data, scalable ingest)
• Applications of machine learning utilizing HPC
• Future research challenges for machine learning at large scale.
• Large scale machine learning applications

Authors are invited to submit full papers with unpublished, original work of 8-12 pages. Submissions will be subject to a double blind peer review process. Submissions will be selected to include both application focused work utilizing ML and HPC and novel methods enabling ML on HPC. All papers should be formatted using the IEEE conference format.https://www.ieee.org/conferences/publishing/templates.html In support of the SC reproducibilty initiative, we also encourage authors to include reproduciblity appendices: https://sc18.supercomputing.org/submit/sc-reproducibility-initiative/ All accepted papers (subject to post-review revisions) will be published in the IEEE Xplore library by IEEE TCHPC. Papers will be submitted through the main SC submissions page https://submissions.supercomputing.org.