Machine learning — also known as cognitive computing — uses predictive analytics to make sense of past data for improved results. It’s a field set to represent a market of more than $20 billion for enterprise software vendors by 2020. Thanks to developments in genetic algorithms and neural networks that mimic the human learning process, machine learning has become increasingly relevant to enterprise applications. It’s hyped to transform forecasting and real-time predictions in detecting anomalies, recommending products or predicting churn. Its applications are robust: whether applied to spotting cybersecurity patterns, sorting produce in agriculture, generating 3D models in construction, or predicting social media trends (to name a few), it will be more and more pervasive over the coming years.

A growing number of predictive analytical models for IoT are being developed around machine learning — it’s timely, given that the era of IoT and big data has unleashed a torrent of unstructured data that is chaotic, complex, and dynamic. Traditional predictive analytics infrastructure based on linear relationships between cause-and-effect variables is becoming increasingly unsuitable. The key value of machine learning in IoT is its ability to counter the butterfly effect, defined as tiny unexpected variations in data behavior within the deluge of big data that has the ability to destabilize predictive models. With big data now beginning to account for a huge chunk of enterprise data, that data is becoming more and more vulnerable to the butterfly effect. The continued growth of IoT as a whole will bring with it infinite data sources in new combinations that make predictions more complex, and machine learning has the ability to make sense of it even in chaotic conditions.

Learn more about what machine learning looks like today from those who are driving its progress at this year’s TiEcon!