Almost everyone today uses some form of machine learning unintentionally.
Several algorithms support the search features of Google, Facebook, or LinkedIn. The same goes for selecting assets on Netflix, Spotify or YouTube. Even when you use your phone for Siri, Okay Google or Amazon Alexa, the voice recognition and search commands are supported by machine learning.
Facebook stopped an “Artificial Intelligence Engine” after the developers discovered that the AI had created its own unique language that they could not understand. Researchers from the ‘Facebook AI Research Lab’ (FAIR) found that the chatbots had deviated from the script and communicated in a new language without human input. Regardless of the perils, this incident clearly reflects the potential of Machine Learning if harnessed with precaution.
Beyond these platforms, ‘Machine Learning’ is a buzzword increasingly used in the location-intelligence domain. However, the first ever algorithms were used over 50 years ago. You may ask what makes machine learning important and relevant now? The answer simply is ‘data’.
2.7 Zettabytes of data exists in the digital universe today
We live in a dynamically changing world feeding a constant stream of structured and unstructured data to unlimited volumes of data storage. If you combine this with computer power and expertise required from people and organisations, the need for machine learning is inevitable.
20%-35% of operating revenue can be lost by businesses due to poor data
The concept of machine learning involves using statistics or mathematical techniques that enable computers to learn without being explicitly programmed.
Hexagon Geospatial, the leading provider of geospatial technology, has been dabbling with machine learning. Their aim is to make this concept available in practice for any geospatial professional.
While various Machine Learning libraries can be found on the internet, most of these have no commotion in the geospatial domain. Beyond the in-depth knowledge on the subject, one must know scripting or programming languages to be able to apply machine learning to geographic datasets.
The Hexagon Geospatial 2018 product release harnesses machine learning and deep learning analysis possible for geographical data. Now available in the ‘Spatial Modeler’, within ERDAS IMAGINE® or GeoMedia®, machine learning enables data to perform multi-class predictions. It would be interesting to see how users adapt to this technology and the effect it has on the output.