Workshop – Hands-On Deep Learning in the Cloud: Fast and Lean Data Science with Tensorflow, Keras, and TPUs

Workshop Description

Training deep learning models used to be a game of patience. Training runs took hours and you needed hundreds of them to tune a model. Well, at least, it gave researchers the time to read new ML papers…

Today, as software engineers add machine learning to their skill sets, they need to work faster because they have products to ship. Making use of the cloud is a big component of that, since it provides nearly unlimited compute resources, but introduces costs that you need to keep an eye on. After all, to leverage the full-scale potential of deep convolutional neural networks, it takes a full-scale solution – for both model training and deployment.

Large-Scale Image Classification

To provide specific hands-on experience, this workshop focuses on image classification applications – culminating with a large-scale problem on real data – by way of convolutional neural networks, which are the most common form of advanced deep learning applied to larger-scale image classification problems.

During this workshop, you will gain hands-on experience training deep learning on Google’s TPUs (Tensor Processing Units).

An in-house Google expert and renowned lecturer will show you the full potential of convolutional neural nets and TPU – without a Ph.D. This workshop will bring you through the full deep learning journey, from the data to the model architecture and model tuning, all the way to deploying models in production.

You will learn to leverage Google Cloud and related tools. With a focus on not wasting time and money, TPUs train faster and at a lower cost. Google’s Colaboratory offers free GPUs and TPUs, as well as notebook sharing and collaboration. And Google’s Machine Learning Engine allows you to run trainings overnight, in parallel, on powerful clusters.

During this workshop day, attendees will gain the following practical hands-on experience:

• Practical steps to deploy TensorFlow and Keras within Google Cloud.
• High-level TensorFlow code using Keras, layers and Datasets.
• Engineering tips, tricks and best practices to build and train neural networks that solve your problems.
• Convolutional neural network architectures for image processing. Convnet basics, convolution filters and how to stack them. Learnings from the Inception model: modules with parallel convolutions, 1×1 convolutions. A simple modern convnet architecture: Squeezenet. Convenets for detection: the YOLO (You Only Look Once) architecture.
• Full-scale model training and serving with Tensorflow’s Keras APIs on Google Cloud ML Engine and Cloud TPUs.
• The application of convolutional neural network training across a real, challenging image classification data set – which will take only minutes to train thanks to the use of TPUs.