Page 38 - RC18-EDGE FALL ALL PAGES - Hi-res

ments and how they envision AI being used to enhance Learning models with Auto-Machine Learning, to a full
a selected business process. The number and type of AI autonomous integrated development environment where
Platforms available today can be daunting. Check out this you can code using with your favorite libraries.
extensive list of the Best AI Platforms Software for 2018.
2
A company’s selection process would evaluate the project TensorFlow
requirements to determine which of the following fea- Open source machine learning framework for
tures must be included: high-performance numerical computation.
Flexible architecture allows easy deployment across a
• Drag-and-drop ability for developers to drag & drop variety of platforms, and from desktops to clusters of
code or algorithms when building models servers to mobile and edge devices.
• Pre-built Algorithms for simpler model development
Amazon SageMaker
• Natural Language Processing service Fully-managed platform that enables developers

• Computer Vision for image recognition services and data scientists to quickly and easily build,
train, and deploy machine learning models at any scale.
• Model Training by supplying large data sets for
training individual models
Bots
• Artificial Neural Networks for users and applications Once the platform has been selected and the specific
that need them functionality determined, the developers may wish to
• Managed Services to assist the users and reduce the employ the use of one or more ‘bots’ (preprogrammed
3
need for infrastructure robots) in their models. The following chart shows a
large sampling of bots that are available for very specific
• Natural Language Generation services
functions within an application.
If a company is considering an AI project, the develop- Conclusion
ment team will most likely request access to one or more So, is AI hype or reality? The growing consumer accep-
of these popular platforms used by businesses in multiple tance of interactive natural language processing assis-
industries: tants such as Alexa, Cortana, Bixby, Siri, and Google will
drive the demand for these tools into the workplace—just
Google Cloud Machine Learning Engine as the consumerization of the iPad did years ago. Models
Uses Googles’ core infrastructure, data analytics that leverage AI structures are becoming the new normal
and machine learning. Secure and fully featured in the development world—an inevitable reality with little
for all enterprises. Committed to open source and indus- or no exaggeration about their benefits. Steven A. Cohen
try leading price-performance. and Matthew W. Granade recently published an article
in the Wall Street Journal titled, “Models Will Run the
Salesforce Einstein World.” In it, they expounded on these points:
4
Built into the Salesforce Platform, Einstein is a
layer of AI that delivers predictions and recom- • There is no shortage of hype about artificial
mendations based on your unique business processes intelligence and big data, but models are the source
and customer data. of the real power behind these tools. A model is a
decision framework in which the logic is derived by
Microsoft Azure Machine Learning algorithm from data, rather than explicitly programmed
A powerfully simple browser-based, visual drag- by a developer or implicitly conveyed via a person’s
and-drop authoring environment where no cod- intuition. The output is a prediction on which a decision
ing is necessary. Go from idea to deployment in a matter can be made. Once created, a model can learn from its
of clicks. successes and failures with speed and sophistication
IBM Watson Studio that humans usually cannot match.
Build and train AI models, and prepare and ana-
lyze data, in a single, integrated environment. • A model-driven business, then, uses models to power the
key decisions in its business process, creating revenue
Deep Cognition streams or cost efficiencies. Building this system requires
Simple, but powerful GUI where you can drag- a mechanism (often software-based) to collect data,
and-drop neural networks and create Deep processes to create models from the data, the models
themselves, and a mechanism (also often software based)
to deliver or act on the suggestions from those models.

36

RC18-EDGE FALL Layout + Mktplce - FINAL.indd 36 10/4/18 3:00 PM