Machine Learning & The Intelligent Edge: The Strategic Option

October 4, 2018

The Strategic Option for Developing and Deploying Algorithms

Similar to cloud computing, machine learning promises to have a profound, transformative impact on business and consumer life. Still in its early stages, machine learning occurs mostly by calling upon centralized resources either on-premise or in the cloud. Unfortunately, this approach restricts companies from capitalizing on the full potential of machine learning.


The drawbacks of centralized machine learning

Whether businesses are transferring their data to the cloud or an on-premise data warehouse, it often takes too long and costs too much to move it to a centralized location. Google’s own analysis of moving data to Google Cloud Platform underscores the gap between the promise of machine learning and the reality where data-transfer issues are concerned. According to Google, it can take days, months, or even years to ship large volumes of data across geographies.

As sensors, machines, and other connected devices generate terabytes and even petabytes of data, it’s simply impractical for companies to continually move data to a central location for analysis. In addition to their time and cost concerns, they must address privacy regulations as data traverses geographies. As a result, many data scientists have to resort to sampling data, severely undermining the ability to fully extract value from their data.


Eliminating the need to move data

Intelligent edge software is making it possible to develop and deploy sophisticated machine learning at the edge, overcoming latency, cost and privacy concerns and the limitations of sampling data. The term “intelligent edge” can be confusing as there are numerous different approaches to edge-based machine learning. For example, in one approach, machine learning happens inside of low cost, low power devices/gateways. In another, it occurs very close to the device, but not within, and provides efficient storage and rapid access to data from an unlimited number of devices. Both approaches offer certain advantages over centralized machine learning, but each is best suited for different scenarios and applications – and can be powerful when used in combination with each other.

Read our latest white paper to learn the pros and cons of each approach and make the best decision as to where to develop and deploy machine learning for your application use cases.


Neil Cohen is the VP of Marketing at Edge Intelligence


Recent Updates

Sign Up for Our Newsletter

  • This field is for validation purposes and should be left unchanged.