Machine Learning: Round Table discussion and demo

Agenda

• Round table discussion

Open source hackers and machine learning enthusiast will meet with an informal panel of Machine Learning experts from Microsoft to discuss how the company is working with developers to help them learn new technologies, deploy innovative solutions and influence the future of features and product. You will have a chance to have an open discussion with

Neil Hutson, GM of Engineering Engagement and Evangelism - https://www.linkedin.com/in/neil-hutson-9a85b9/

Micheleen Harris, Machine Learning Lead - https://www.linkedin.com/in/neil-hutson-9a85b9/

Hannah Krager, Machine Learning Lead - https://www.linkedin.com/in/hannahkrager/

Nigel Parker, Chief Engineer, CSE Asia - https://www.linkedin.com/in/nzigel/

• Demo: Large Scale DNN Training with Tensorflow, Horovod and Keras

In this lightning talk we’ll do a quick lap around options for training large scale Tensorflow models.We’ll discuss patterns of scale out including the ScatterReduce-AllGather MPI communication approach.
In this session you’ll see how to train across multiple GPUs and Multiple machines running multiple GPUs in the Azure cloud.

Chris Auld - Engineering Manager for the Microsoft CSE team covering South East Asia https://www.linkedin.com/in/cauld/

• AI on the Edge (Open Neural Network Exchange format)

There is a trend to move machine learning models closer to where they are being executed. In this session Nigel will demonstrate ways to build image classification models that can be run disconnected from the Internet on Android, iOS and Windows ML using both customvision.ai and other transfer learning techniques. Nigel will also talk about Windows ML for offline hardware accelerated execution of ONNX.ai (Open Neural Network Exchange format) machine learning models in the next version of windows