Numenta

Home of the HTM Community

Welcome to Numenta.org, home of Numenta’s HTM community and open source projects. If you want to learn about Numenta the company visit Numenta.com.

Machine Intelligence Starts Here

Hierarchical Temporal Memory is a foundational technology for the future of machine intelligence based upon the biology of the neocortex. Because Numenta is committed to making this technology accessible to everyone, all HTM software and ongoing research is open source. This allows you to work with our technology in whatever way works best for you ‒ learn about the theory, dive into the source code, or start your own implementation. Some of our community members have written their own versions of HTM systems in other languages and platforms. Others have created detailed visualizations, experiments, and applications.

The neocortex is a logical system that we’ll understand fully in time. HTM theory reflects our current understanding of how the neocortex works, and HTM code reduces that theory to practice. HTM is continually being updated as we learn more about the brain. We believe HTM will play a critical role in the creation of truly intelligent machines.

Join us!

There are many ways for you to get involved. Come discuss HTM theory with us on our forums, or share your HTM application on our Github organization. If you’re new to HTM, a great place to learn is in HTM School. We hope you’ll join our community and help us discover the future of machine intelligence.

Hierarchical Temporal Memory

Hierarchical Temporal Memory (HTM) is a biologically-constrained theory of intelligence originally described in the book On Intelligence. HTM is not a Deep Learning or Machine Learning technology. It is a machine intelligence framework strictly based on neuroscience and the physiology and interaction of pyramidal neurons in the neocortex of the mammalian brain.

If you’re interested in learning more about HTM, visit our educational series HTM School or browse through the topics on HTM Forum. You can also find a collection of research papers at https://numenta.com/resources/papers/ and a living book authored by Numenta researchers and engineers that documents HTM called Biological and Machine Intelligence (BAMI).

Our community is an eclectic collection of researchers, scientists, hobbyists, and hackers interested in building biologically-inspired intelligent systems with Hierarchical Temporal Memory (HTM).

Sponsored by Numenta and supported by a full-time employee dedicated to fostering and growing the community, we strive to create an inclusive environment. Our community exemplifies the type of people drawn toward new technology, and it shows in our curiosity and open-mindedness.

To get involved in our community, join HTM Forum. You can login with your Google, Facebook, or Twitter account, or by creating a new account with your email address. This is the best place to ask questions, search for answers, or just interact with others working on similar problems. We aim to be a helpful and welcoming community, and we hope you'll join us.

While many of our interactions are virtual, we do hold community events like meetups, hackathons, and workshops throughout the year. All our live events are scheduled via Meetup. Please join our Meetup Group for notifications of upcoming events, all of which are free and open to anyone. Please read our code of conduct before attending events.

For recordings of previous events, please see our YouTube channel. You may also be interested in Numenta's company events and archive.

In addition to making all of Numenta’s production and research code open source, our community produces and shares HTM implementations and applications, commonly posting their projects on HTM Forum. Our community hosts many of these projects on the HTM Community Github organization. We have HTM implementations in languages like Python, C++, Java, Clojure, Go, and JavaScript.

We are always looking for community members that can help others build out HTM applications and experiments. If you are interested in helping or providing consulting services, please leave us a note on the HTM Hackers Forum and introduce yourself.

---

I’m pleased to announce that we have a new commercial licensee, Intelletic Corporation (IC), a financial startup developing an AI platform for autonomous trading of futures and other financial assets. Using Numenta’s Hierarchical Temporal Memory (HTM) technology, IC has built a proprietary Cortical Learning Model (CLM) that locates trade opportunities in financial time series and executes those trades in real time with less risk and greater return than a human discretionary trader. The team at IC has been working with HTM in open source for several years, and we’re excited to see their efforts progress to a commercial endeavor. With excellent back-test results, IC is currently raising Series A venture funding to operationalize and optimize their trading platform. Visit www.intelletic.com to learn more about the company.

Our research team has been actively exploring our most recent brain theory development: the potential role of grid cells in the neocortex. Last month, our Co-founder Jeff Hawkins spoke on the UC Berkeley campus at the Computational Theories of the Brain Workshop. In his talk, “Does the neocortex use grid cell-like mechanisms to learn the structure of objects?” Jeff presented our most recent research discoveries by walking through a proposal of how cortical columns pair sensory input and location to learn the structure of objects. I invite you to watch the video of his talk and post any questions or comments on our open source discussion board thread on the topic.

Research Engineer Marcus Lewis presented a poster titled “Using Grid Cells for Coordinate Transforms” at the Grid Cell Meeting in London May 21-22. In the poster, Marcus walks through a detailed mechanism that could explain how grid cells in the cortex enable object recognition.

In partner news, Grok has been busy building their latest version, Grok 3.0, scheduled for a limited release at the end of the month. Version 3.0 includes the addition of a microservices-based backbone capable of scaling to accommodate very large deployments, thus allowing real-time monitoring of data streams to increasingly improve incident detection. You can be one of the first to preview the new version by requesting a demo at https://www.grokstream.com/demo/.

Cortical.io recently released a new version of their free tool, Iris, which allows you to perform intelligent text analysis, information extraction and text-data comparison. The latest release allows you to enter a few examples and classify your text in 8 different languages. You can learn more and download Iris here.

Francisco Webber, Cortical.io Co-founder, will be speaking on May 29 at the Frankfurt Finance Summit about how financial institutions have used the Cortical.io Contract Intelligence Engine to improve information extraction for legal documents.

Lastly, we have a new piece on our blog titled “The 2018 Machine Intelligence Landscape,” which looks at what progress has been made in the various approaches to machine intelligence - from Classic AI too Deep Learning to the Biological Neural Network. The piece is an update to a blog our co-founders Jeff Hawkins and Donna Dubinsky wrote in 2016. Check it out to see why we believe now more than ever that brain theory provides the roadmap to general machine intelligence.

Thank you for continuing to follow Numenta.

Christy Maver
VP of Marketing