English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

We present ease.ml, a declarative machine learning service platform we builtto support more than ten research groups outside the computer sciencedepartments at ETH Zurich for their machine learning needs. With ease.ml, auser defines the high-level schema of a machine learning application andsubmits the task via a Web interface. The system automatically deals with therest, such as model selection and data movement. In this paper, we describe theease.ml architecture and focus on a novel technical problem introduced byease.ml regarding resource allocation. We ask, as a "service provider" thatmanages a shared cluster of machines among all our users running machinelearning workloads, what is the resource allocation strategy that maximizes theglobal satisfaction of all our users?  Resource allocation is a critical yet subtle issue in this multi-tenantscenario, as we have to balance between efficiency and fairness. We firstformalize the problem that we call multi-tenant model selection, aiming forminimizing the total regret of all users running automatic model selectiontasks. We then develop a novel algorithm that combines multi-armed bandits withBayesian optimization and prove a regret bound under the multi-tenant setting.Finally, we report our evaluation of ease.ml on synthetic data and on oneservice we are providing to our users, namely, image classification with deepneural networks. Our experimental evaluation results show that our proposedsolution can be up to 9.8x faster in achieving the same global quality for allusers as the two popular heuristics used by our users before ease.ml.

Ease.ml: Towards Multi-tenant Resource Sharing for Machine Learning Workloads