2

Is there a way to increase the processing power of the Azure ML? I've deployed a neural network on a huge dataset (8000+ retina images, and Azure is taking an impossible amount of time to run the programme. Is it possible to deploy the ML workspace from a Virtual Machine, so that I can leverage increased processing speeds? Help!!

Mike Wise
  • 22,131
  • 8
  • 81
  • 104
Saptarshi
  • 21
  • 2

1 Answers1

1

There's no way to deploy Azure ML workspace in a virtual machine. But Microsoft provides a VM specifically designed for data science related tasks. (https://azure.microsoft.com/en-us/documentation/articles/machine-learning-data-science-provision-vm/) You can customize the resources there.

On the other hand if you upgrade the AML Studio to standard tier, the execution may perform using multiple nodes. It will increase the performance of the experiment drastically compared to the single node execution in AML Studio free tier.

Haritha Thilakarathne
  • 880
  • 8
  • 19
  • Haritha, I have already switched to a Standard Tier. But the performance is still too slow even on simple Linear Regression – Saqib Ali Apr 24 '17 at 17:29
  • i also upgraded to the standard tier and there was zero increase in the speed of training :-( – blissweb Sep 20 '18 at 01:42