I'm going to do machine learning with GCP! Ok, but how?

There are tons of features on GCP. But what should you be using if you want to work on an ML project? We’ll touch on the two main ML activities, (1) using off the shelf AutoML services, and (2) spinning up your own ML project that can take advantage of the awesome horsepower of the Google Cloud.

This presentation will provide a broad overview of how someone can utilize the Google Cloud Platform when doing a Machine Learning project. The two main areas of focus will be (1) off-the-shelf features (like Vision and Natural Language) and (2) using general features (like the virtual machine, Storage, and Dataflow) to spin up your own machine learning models.

In general, when doing a machine learning project, you try to break a project up into pieces (rather than trying to do one big training model that does everything). For each piece, you determine if you can use an off-the-shelf type of service. For example, you might use the AutoML Vision service on GCP if you want to identify where a face is in an image as one of the steps. Once you’ve exhausted all your options for off-the-shelf services, you will want to address the remaining pieces with your own ML models. This presentation will step you through this process, covering the off-the-shelf services you may be able to utilize, and then moving on to setting up a Virtual Machine to build your own Machine Learning models with the typical tools used in the field (i.e. Jupyter, Cuda, Torchvision, etc.).

I inquired with contacts at Google to see if they will provide extra GCP related promotional offerings. Right now, anyone can get a $300 credit towards, so I’ll mention that. But hoping for more.