How machine learning is done on the cloud by Amazon, Microsoft, IBM, and Google

Any cloud must meet an ever-increasing set of qualities in order to be considered seriously. Given that machine learning is currently offered by all of the main cloud providers, it appears to be on that list.  What is different is how they go about achieving it. There are several approaches besides the "curated API vs. open-ended algorithm marketplace": "everything and then some vs. just enough." This is how the four major cloud providers, Amazon, Google, IBM, and Microsoft, compare to one another in terms of machine learning.

IBM: Watson at the helm, turning the ship

The questions multiplied when IBM first revealed it would make its Watson AI system available as a consumable service. How might it appear? In what way would it be eaten? Primarily, to what extent would it aid IBM's endeavor to reposition itself as a dominant player in the cloud space?  After around two years, IBM has launched a number of machine learning-powered services on its Bluemix PaaS, including systems for sentiment and tone analysis, language analysis, image identification, weather prediction, and more.