This follows Microsoft's announcement in February when it launched the Azure Machine Learning Platform, and just last month IBM announced that it had bought AlchemyAPI to boost its—wait for it—machine learning capabilities on its Watson cloud service on Bluemix.
What all of these services have in common is that they involve processing large amounts of data in the cloud and then building applications that make use of the data.
They all have a similar process of ingesting the data, building a model, and making predictions. Those predictions could involve sales churn or product recommendation engines, not unlike the kind you've seen for years on Amazon.com.
Big data is a big deal these days for a reason. Taking large amounts of data, processing it, and finding answers to questions is something every company wants to be able to do. Up until now it has been mostly a backward-looking exercise. Given the data I have, what happened? Now it's moving more to real time, but where it really gets useful is in making predictions.
Machine learning techniques moves to the cloud
The cloud is uniquely suited to a machine learning technique, something Google obviously recognized early and its competitors have begun to catch up on. The thing about this type of data analysis is that it typically involves large amounts data and requires significant computing power to process and analyze.
That's where the cloud comes in. It can scale up to deal with the influx of data and give you cheap storage and as much computing power as you need for as long as you need it. When you're done, you can store your source data in a cheaper long-term storage option, and you'll only be billed for the compute resources you use and no more.
When the project is over, you can release those resources and then draw on them again whenever you need them. It's all the beauty of the cloud in one compact example. The cloud vendors noticed that if they could solve this problem, they could generate a lot more cloud business, especially as more companies are using data on a regular basis.
The thinking behind of all these solutions is that they provide a way to feed their cloud service business and offer a range of services that can help generate a significant amount of money. The only surprising part of this is that it took AWS so long to figure it out.
To be fair, companies were processing machine learning projects on the AWS platform before this week's announcement, It was just challenging to do it without a specific set of tools and APIs. AWS answered that this week and put itself on par with its competitors in the process.
Now that everyone offers a machine learning tool in the cloud, it's up to you to decide which one you want to use.
Photo by xdxd_vs_xdxd on Flickr. Used under CC by SA-2.0 license.