Microsoft has announced the launch of Azure Batch AI training at its Build 2017 conference. The service lets users train deep neural networks on Microsoft’s Azure cloud computing platform. While currently available in private preview, it’s full potential will be exploited when it goes public.
So what is batch AI training? It is basically a managed training service that allows developers to train their models without having to worry about what goes into the training process. Think of it this way — you decide upon the model and let Microsoft do the heavy lifting. The company wants to provide developers with a way to train their models without having to worry about the technical nuances involved.
With this service, developers will be able to run their models and workloads on multiple CPUs and GPUs, which is something that will be highly appreciated by those with complex workloads.
What’s more, the service supports any frameworks ranging from Google’s TensorFlow, Microsoft’s own Cognitive Toolkit or Caffe and so on and so forth. The environment that you want to train your model in will be entirely at your discretion and this will also decide how much they will need to shell out once the service moves out of private preview and into the public mode.
Considering the growth of AI and a need for reliable machine learning models, this could turn into a steady money spinner for the Redmond giant.