Christy Maver

Director of Marketing at Numenta


Scott Purdy

Director of Engineering at Numenta

Machine Intelligence with Streaming Data
About the Webinar

Across every industry, we are seeing an exponential increase in the availability of streaming, time-series data. The real-time detection of anomalies has significant practical application. Finding anomalies in such data can be very difficult, given the need to process data in real time, and learn while simultaneously making predictions. With the increasing variety of streaming data sources, automated deployment—without manual parameter tuning—is also becoming important.

Numenta’s online sequence memory algorithm, called Hierarchical Temporal Memory (HTM), has been used to detect anomalies in IT monitoring, human behavior, the stock market, geospatial data, and more. This webinar will introduce this novel technique, demonstrate its broad applicability, and cover performance details from a published benchmark designed for real-time anomaly detection.

About Presenters

As Director of Marketing, Christy Maver is responsible for all aspects of marketing and communications at Numenta. Christy has 17 years of experience with technology marketing, including 13 years in various roles at IBM. She holds a BA in Economics from Princeton University. Follow Christy on Twitter @cdmaver


Scott Purdy is Director of Engineering at Numenta. He works with the research team, collaborates on the open source NuPIC project, and manages the backend teams. Prior to Numenta, Scott worked in engineering at Google and earned his B.S and M.Eng. from Cornell University's College of Engineering. He is a native of the Pacific Northwest and has called San Francisco home for the last five years.

About Numenta

At Numenta, we are tackling one of the most important scientific challenges of all time: reverse engineering the neocortex. Studying how the brain works helps us understand the principles of intelligence. We apply our discoveries to one of the most important technology challenges of all time: creating intelligent machines. We believe that understanding how the neocortex works is the fastest path to machine intelligence, and creating intelligent machines is important for the continued success of humankind.

Based on a wealth of neuroscience evidence we have created a technology called HTM (Hierarchical Temporal Memory) that models key aspects of neocortical learning. HTM is ideally suited for classifying, making predictions, and detecting anomalies in streaming data. Unlike most other machine learning methods, HTM learns time-based patterns in data on a continuous basis. All of our HTM algorithm and application code is available for free under the AGPLv3 open source license and our scientific work is available on this website and in peer-reviewed publications. We have created demonstration HTM applications in several fields such as IT monitoring, detecting unusual human behavior, geospatial tracking, and understanding natural language and we are confident that many additional applications will be created in the future.