Google Cloud ML Brings a Unique Angle to the Machine Learning Cloud
Google Cloud Machine Learning (Cloud ML) is a new addition to the cloud machine learning market. While PaaS leaders such as AWS and Azure have released their ML stacks more than a year ago, Google Cloud certainly has taken its time to enter the space. However, the first version of the cloud ML service already provides some unique differentiators that can push it to a leadership position within that segment of the ML market.
Conceptually, Cloud ML can be seen as a native cloud service for hosting TensorFlow applications. Cloud ML provides an architecture based on concepts such as projects, models, versions and jobs that enables the implementation of really sophisticated ML applications.
Competition is Intense
Cloud ML is entering a very competitive market with incumbents such as Amazon and Microsoft leading the charge with very innovative offerings. Azure ML is Microsoft’s ML cloud service which offers some interesting technical capabilities such as visual model building and strong support fo rR and Python ML scripts. Similarly, AWS ML excels at the simplicity of its programming model and the strong support for advanced statistical methods.
Despite the initial traction of both Azure ML and AWS ML bot platforms have well-known limitations that constrains its usage to relatively simple ML scenarions. Today, is really complicated to build a sophisticated ML solutions using those services exclusively due to limitations in areas such as extensibility. support for custom algorithms, integration with on-premise data sources, etc. Google Cloud ML provides a unique model that addresses some of those limitation without sacrificing the simplicity of the programming model.
5 Factors that can Make Cloud ML a Winner
The TensorFlow Factor
TensorFlow is one of the most popular open source deep learning frameworks in the market. by leveraging Tensorflow, cloud ML allows developers to implement really sophisticated and highly extensible ML programs without investing in complex infrastructures.
The DeepMind Factor
Google’s DeepMind recently adopted TensorFlow as its underlying stack. The complexity of the AI problems that DeepMind is attacking is likely to translate into improvements for the TensorFlow stack and subsequently Cloud ML.
The AI API Factor
Google Cloud continues making steady progress on the AI space. The recent releases of capabilities such as natural language processing (MLP) APIs or Speech APIs are an example of the rapid growth in the Google Cloud AI ecosystem. Even though competitors such as Azure offer the same levels of AI and ML services as part of its platforms, other PaaS such as AWS and Bluemix are still trying to structure a cohesive platform for both types of capabilities.
The Hybrid Factor
As a side effect of using Tensorflow, Cloud ML customers can develope solutions completely on-premisee and deploy it and scale it in cloud environments. That llevelof symmetry is aabsentof other cloud ml stacks.
The Google Cloud Factor
Finally Cloud ML integration with unique Google Cloud services such as Datalab (self-service data science) or Bigquery enables and implementation of really robust ML solutions.
Google Cloud ML is a very young but also exciting addition to the cloud machine learning ecosystem. While Amazon, Microsoft and IBM still lead the field, Google Cloud ML has the potential of becoming a relevant solution in the space in a very short time.