I’m excited to announce that we’ve launched a new podcast series called Numenta On Intelligence. This monthly podcast will feature conversations about what it means to figure out how the brain works and understand the principles of biological intelligence, and why it’s such an important problem to solve. Whether you’re an algorithms expert, brain science enthusiast or simply curious about how your brain works, I encourage you to listen in. You can subscribe on iTunes, Stitcher, Google Play, or wherever you download your favorite shows.
This month also kicks off a steady stream of events for Numenta, starting with CNS 2018 (Organization for Computational Neurosciences). We will have four different activities at the event that cover our foundational research on the neuron model and sparse distributed representations (SDRs), as well as our newer research on location signals and grid cells. Our VP of Research, Subutai Ahmad, will present two workshops: The Predictive Neuron: How Active Dendrites Enable Spatiotemporal Computation in the Neocortex and Locations in the Neocortex: A Theory of Sensorimotor Prediction Using Cortical Grid Cells. Members of our research team will also join Subutai in presenting two posters: Robust Dendritic Computations with Sparse Distributed Representations and Unsupervised Learning of Relative Landmark Locations Using Grid Cells. Full details of the workshops and posters are available on our CNS 2018 event page. As always, we’ll make the materials available on our website after the event.
For those of you who have followed our work’s applicability to anomaly detection for streaming data, we have a couple new items that may interest you. First is a new paper published by a team of researchers at the University of Melbourne Australia, “Detecting performance anomalies in scientific workflows using hierarchical temporal memory.” As the title suggests, the paper proposes an anomaly detection framework for scientific workflows that is based on HTM. The team also uses the Numenta Anomaly Benchmark (NAB) scoring code and methodology for to compare HTM performance against other algorithms. If you’re not familiar with NAB, it is an open source benchmark we designed specifically for time-series data that gives credit to early detection and adjusting to changed patterns. We are pleased to see an independent validation of HTM, as well as the application of HTM to a new domain, in a peer-reviewed journal.
Second is a blog we published on a new algorithm that we came across in an AWS post and decided to run it through NAB. Read the blog to see where the algorithm placed on the NAB leaderboard, how we arrived at the results, and how you can replicate the results to see them for yourself.
In partner news, Cortical.io, a Vienna-based Natural Language Understanding company that creates solutions based on how the brain processes information, has published two new case studies:
- Contract Intelligence: How a Big Four accounting firm reduced review time of lease agreements by 80%
- Support Intelligence: Implementation of the Cortical.io Support Intelligence Engine at a Fortune 100 company
Cortical.io was also featured in a recent TWiML&AI (This Week in Machine Learning & AI) podcast episode.
Thank you for continuing to follow Numenta.