AWS Summit San Francisco: AI/ML recap

Outside of re:Invent, the San Francisco summit is one of the bigger events when it comes to announcing AWS new services.

The tradition was upheld again this year. Let’s take a quick look at the different AI/ML announcements.

Illustration for AWS Summit San Francisco: AI/ML recap

1 — Machine Translation: Amazon Translate

Available in preview at re:Invent 2017, Amazon Translate is now generally available in the US East (N. Virginia), US East (Ohio), US West (Oregon), and EU (Ireland) regions.

Need a quick look at Translate? This post will tell you what you need to know.

We also announced that new languages would be supported in the coming months: Japanese, Russian, Italian, Traditional Chinese, Turkish, and Czech.

2 — Speech to Text: Amazon Transcribe

Available in preview at re:Invent 2017, Amazon Transcribe is also generally available in the US East (N. Virginia), US East (Ohio), US West (Oregon), and EU (Ireland) regions.

Need a quick look at Transcribe? This post will tell you what you need to know.

3 — Machine Learning: Amazon SageMaker

The TensorFlow and Apache MXNet containers using for training and prediction by Amazon SageMaker have been open-sourced!

Need a quick look at SageMaker? This video will tell you what you need to know.
aws/sagemaker-tensorflow-containers
sagemaker-tensorflow-containers - This support code is used for making the TensorFlow framework run on Amazon…github.com
aws/sagemaker-mxnet-containers
sagemaker-mxnet-containers - This support code is used for making the MXNet framework run on Amazon SageMaker.github.com

This is great news for all SageMaker users, who are now able to:

  • build and customize these containers for their own needs,
  • run them locally or on-premise for development and testing purposes,

New notebook instance sizes have been introduced. You can now go as big as ml.m4.16xlarge and ml.p3.16xlarge, which will help you run heavier data pre-processing on the notebook instance itself.

4 — Machine Learning at the Edge: AWS Greengrass ML Inference

Available in preview at re:Invent 2017, AWS Greengrass ML Inference is now generally available, in the same regions as AWS Greengrass: US East (N. Virginia), US West (Oregon), EU (Frankfurt) APAC (Tokyo) and APAC (Sydney) regions.

5— Apache MXNet: Model Server

The latest version of the MXNet Model Server (v0.3) now lets you use pre-built CPU and GPU containers stored on the Docker Hub.

Happy to answer any questions. Feel free to reach out on Twitter.