Advancing Machine Intelligence with Neuroscience

Numenta’s deep experience in theoretical neuroscience research has led to tremendous discoveries about how the brain works. We are applying the principles of real intelligence to the world of artificial intelligence to create machines that can understand the world and add great value to humanity.

The brain is the only example we have of an intelligent system. Understanding how it works and what makes it intelligent allows us to build machines that work on the same principles. We believe that a neuroscience-based approach is the fastest path to creating general intelligence, and we’ve laid out a roadmap to get there.

Near term, we are applying our model to existing deep neural networks in order to enable systems that are more brain-like, such as systems that are robust to noise and can learn continuously. Long term, we will continue to incorporate our neuroscience principles with the ultimate objective of creating intelligent sensorimotor systems that can learn, plan and act.

Underlying our machine intelligence work is scientific research that demonstrates how the brain create intelligence. In October 2018 we released a major new theory for intelligence and cortical computation. We have several resources available for you to learn more.

What We Do

We are a team of scientists and engineers applying neuroscience principles to machine intelligence research. Our neuroscience research code is publicly disclosed and available in open source.

Our Progress

Numenta has developed a major theory of intelligence and how the brain works called The Thousand Brains Theory of Intelligence, and we’re now exploring how to incorporate key principles of the theory to the field of machine intelligence.

What Makes Us Unique

We are one of the few teams that has developed large scale theories of the brain that are biologically constrained, testable, and implemented in software. We believe The Thousand Brains Theory will be foundational to the creation of truly intelligent systems.

How Our Research Will Impact the Future of Machine Intelligence

Over the past decade, machine learning and AI have succeeded at performing many tasks that were unimaginable earlier. Problems such as identifying a cat in an image or recognizing someone’s speech are now routine in our lives. Yet, the way these technologies accomplish the tasks is fundamentally different than how our brain accomplishes the same tasks. For example, a machine learning system that recognizes a cat may be trained with over a million labeled training images in order to reliably identify a cat, whereas a human child learns about a cat with just a handful of examples.  Moreover, that child knows that the cat may purr, the cat may scratch and the cat can jump high, whereas all the machine knows is what a cat looks like.  The machine doesn’t understand cat; the child does.

The limitations faced by today’s AI systems are generally agreed in the machine learning community.  Such systems require enormous time and resource to train, are brittle to noise, cannot learn continuously, and do not generalize but rather accomplish narrow, specific goals.  We are still a long way from a robot performing tasks that a child can handle easily.

We do not believe it is possible to overcome these limitations simply by proceeding down the same path with more data and more power.  A new approach is needed and we believe that approach must incorporate what we have learned from the brain.

We have a robust road map that applies the principles of neuroscience to put us on the path towards machine intelligence.  In the near term, we are applying our model to existing CNN and AI models in order to enable systems that are more brain-like, such as systems that are robust to noise and can learn continuously.  Long term, we will continue to build out more pieces of our model with the ultimate objective of creating intelligent sensorimotor systems that can learn, plan and act.

Play Video

Video: Why Brains Matter (02:01)

Close Section

Our neuroscience research has uncovered a number of core principles that are not reflected in today’s machine learning systems. These principles have the promise to solve many of the known problems today’s machine learning and AI systems face. By incorporating these principles, we can overcome today’s limitations and build tomorrow’s intelligent machines. Below are some of the topics we are currently researching:

Current Research Projects:

Robustness

Our neuroscience research has shown that sparse representations are more robust and stable than dense representations. We have developed a cortically inspired sparse algorithm that can be applied to deep learning networks trained through backpropagation. These networks have both sparse activations and sparse connections. Sparse networks achieve accuracy competitive with the state of the art dense models, but are significantly more robust to noise.

Dynamic sparse networks

In the brain, cortical networks are sparsely connected and extremely dynamic. As many as 30% of the connections in the neocortex turn over every few days. We are investigating ways to create highly sparse networks that learn their structure dynamically through training.

Performance improvements in sparse networks

Representations in the brain are highly sparse, resulting in an extremely efficient system. For machine learning, sparsity also offers the promise of significant computational benefits, but most hardware architectures are not optimized for extreme sparsity. These limitations have hindered research into sparse models. Along with our hardware partners, we are developing methods for dramatically improving the computational efficiency of sparse neural networks.

Continuous learning

Truly intelligent machines must have the capability to learn and adapt continuously, a property that is absent in today’s deep learning systems. In the brain, sparse representations plus a more complex neuron model, enables us to continuously learn new patterns in an unsupervised manner. We have shown that incorporating these ideas into artificial neural systems can enable systems that learn continuously from streaming data without any manual interventions.

Future Research Projects:

Long term, our focus is on creating truly intelligent machines that understand the world.  Future research projects may include:

  • Learning with much smaller training sets
  • Improving generalization
  • Building integrated sensorimotor systems that can plan, act and learn

Resources

Papers

Ahmad, S., Scheinkman, L., (2019) How Can We Be So Dense? The Benefits of Using Highly Sparse Representations

Cui, S. Ahmad, & J. Hawkins, (2016), Continuous Online Sequence Learning with an Unsupervised Neural Network Model

Ahmad, & J. Hawkins, (2016),. How Do Neurons Operate on Sparse Distributed Representations? A Mathematical Theory of Sparsity, Neurons and Active Dendrites

Cui, S. Ahmad & J. Hawkins (2017) The HTM Spatial Pooler—A Neocortical Algorithm for Online Sparse Distributed Coding

Videos

The Thousand Brains Theory: A Framework for Understanding the Neocortex and Building Intelligent Machines

Sparsity In The Neocortex

Play Video

Video: Applying The Thousand Brains Theory to Machine Intelligence

Close Section

Our Goal

Vernon Mountcastle  proposed that because every part of the neocortex has the same complex circuitry, then every part is doing the same thing. Therefore, if we can understand a cortical column, we will understand the neocortex.

Through our focus on cortical theory, we seek to understand the complex circuitry of a cortical column. We want to understand the function and operation of the laminar circuits in the brain, including what the neurons are doing, what the layers are doing, how they interact, and how a cortical column works.

Video: Key Discoveries in Understanding How the Brain Works
Play Video

Video: Key Discoveries in Understanding 
How the Brain Works (4:33)

Our Approach

We are often asked how we do neuroscience research at Numenta and what it means to focus on cortical theory. We read many neuroscience papers and published results to get a deep understanding of neuroanatomy, neurophysiology, and cortical function. We propose theories about how the detailed architecture of the neocortex implements these functions. We ensure our work satisfies biological constraints and test our theories via simulation. Since we do not do any wet lab work, we collaborate with experimentalists that run investigations and give us feedback. We also host scientists through our Visiting Scholar Program.

We document our research in several ways, including peer-reviewed journal papers, conference proceedings, research reports, and invited talks. In addition, we place our daily research commits in an open source project  and answer questions about our research posted on our forums . We strive to be completely open in everything we do.

Cortical Theory Our Approach

Our Theories

Although there is much to do, we have made significant progress on filling in the pieces of the common cortical circuit that underlines all of intelligence. We have proposed a novel theory and broad framework for understanding what the neocortex does and how it does it. We have made discoveries on how neurons make predictions, the role of dendritic spikes in cortical processing, how cortical layers learn sequences, and how cortical columns learn to model objects through movement.

Cortical Theory Our Theory

Our Research Focus Areas

Cortical Columns

Building on Mountcastle’s proposal that the neocortex is made up of nearly identical cortical columns, we are working on filling in the pieces of the cortical circuit – understanding each layer, what the neurons are doing, and how a cortical column works. As part of this research, we have proposed a framework, consistent with anatomical and physiological evidence, that explains how cortical columns function. The framework is based on how cortical columns learn sensorimotor models of the world by combining sensory inputs with location signals.

Find more information on cortical columns.

Cortical Theory Cortical Columns

Sequence Learning

Memory and recall of sequences is an essential component of inference and behavior. We believe that sequence memory is occurring in multiple layers of the neocortex. We’ve shown how a layer of pyramidal neurons with active dendrites, arranged in mini-columns, will learn transitions of patterns and form a robust sequence memory.

Find more information on sequence learning and prediction in cortex.

Cortical Theory - Sequence Learning

Sparse Distributed Representations

Sparse Distributed Representations (SDRs) are a foundational aspect of all of our theories. Everywhere in the neocortex, information is represented by distributed and sparsely active sets of neurons. We have shown through mathematical analysis and simulation that SDRs enable semantic generalization and robustness.

Find more information on sparse distributed representations.

Cortical Theory - Sparse Distributed Representations

Close Section

We have a growing collection of published peer-reviewed papers, supplemental white papers and research manuscripts. You can search our publications by category or by year. Some are currently under review at journals/conferences, but we have made all manuscripts freely available on preprint sites, such as arXiv or bioRxiv. Our goal is to document all of our discoveries in scientific journals.

  • Research papers • View and download the papers we have published and posted on preprint servers.
  • Conference posters • View and download the posters we have presented at academic events.
  • Outside research • See how other scientists have analyzed our work in these papers that feature our research and technology.
Numenta Research Publications

Close Section

HTM

Based on a wealth of neuroscience evidence, we have created HTM (Hierarchical Temporal Memory), a technology that is not just biologically inspired. It’s biologically constrained. When applied to computers, HTM algorithms are well suited for prediction, anomaly detection and ultimately sensorimotor applications. We believe this technology will be the foundation for the next wave of computing.

At the core of HTM are learning algorithms that can store, learn, infer and recall high-order sequences. Unlike most other machine learning methods, HTM algorithms learn time-based patterns in unlabeled data on a continuous basis. They are robust to noise, and high capacity, meaning they can learn multiple patterns simultaneously.

HTM algorithms work best with data that meets the following characteristics:

  • Streaming data rather than batch data files
  • Data with time-based patterns
  • Many individual data sources where hand crafting separate models is impractical
  • Subtle patterns that can’t always be seen by humans
  • Data for which simple techniques such as thresholds yield substantial false positives and false negatives

Our technology has been tested and implemented in software, all of which is developed with best practices and suitable for deploying in commercial applications.

Video: Intro to Our Technology
Play Video

Video: Intro to Our Technology

HTM Studio

HTM Studio is a free, desktop tool that lets you find real-time anomalies in your streaming data without having to program, code or set parameters.

Download HTM Studio and try it for yourself.

HTM Studio

Open Source

Because we want our technology to be broadly adopted, we make it widely accessible in an open source project. There you’ll find our algorithms, source code, and an active discussion forum  with HTM community members covering a variety of topics.

You’ll also find our latest work with HTM applied to today’s machine learning platforms. For example, we’ve created libraries to create sparse Deep Learning Networks in nupic.torch. Our community has created an assortment of HTM implementations, experiments, and integrations available for study and use.

If you are interested in seeing, developing or working with our technology, we invite you to visit our HTM Community website .

Anyone is welcome to use our technology for free, under the AGPLv3 open source license . In addition, we have created a separate, trial license without commercial rights  for those individuals or organizations who are unable to use the AGPLv3 license. For more on our licenses, see the Licensing & Partners section section.

Numenta Open Source Hackathon Event

Close Section

To help you learn about our theory and technology, we have created a number of videos, podcasts and educational resources. They are designed for anyone who wants to learn more about our cortical theory and HTM technology.

  • Videos From keynote presentations to invited talks to cortical animations, view our library of videos to see our research developments firsthand.
  • Numenta On Intelligence Podcast Our podcast is about intelligence – how it works in the brain, what the key principles are, and how to apply those principles to machine learning systems.
  • Other Podcasts Occasionally, our CEO Donna Dubinsky and Co-Founder Jeff Hawkins appear as guests on other podcasts.
  • HTM School This YouTube series is designed to educate the general public about Hierarchical Temporal Memory (HTM). Each 10-15 minute episode dives into a particular topic.
  • HTM Forum • If you’re interested in discussing our work, or developing and implementing our technology, join our open source discussion forum.
  • Legacy Applications • Early example applications of HTM technology focused on anomaly detection for streaming data.
  • Numenta Anomaly Benchmark (NAB) • We created NAB in order to be able to measure and compare results from algorithms designed to find anomalies in streaming data.
  • Biological and Machine Intelligence (BAMI) This living book (Biological And Machine Intelligence) documents our framework for both biological and machine intelligence.
  • On Intelligence Read the book where Jeff first shared many of the core concepts of our theories. Intended for a lay scientific audience, On Intelligence provides a nice introduction to anyone interested in understanding how the human brain works and what intelligence is.
The Hard Unsolved Problems in HTM Theory
Play Video

Video: The Hard Unsolved Problems in HTM Theory (1:00:00)

Close Section

Numenta’s business strategy and approach to intellectual property (IP) is to create an active research community as well as to enable strong commercial opportunities. To that end, we follow these general principles:

  1. Transparency. We openly publish our scientific findings, software, intellectual property, and business strategy. There are no hidden agendas.
  2. Scientific Use. Anyone can freely use our software and intellectual property for non-commercial purposes.
  3. Commercial Deployment. We make reference software available (including relevant patents) for free under an AGPL license. If the AGPL license is not a good fit, or if you are interested in using our intellectual property without our software in a commercial application, we offer an IP license that enables commercial use. Before entering into a commercial license, we require a validation checkpoint, which can be done under an AGPL or Validation License.

Learn more about Numenta’s business principles.

Summary of Numenta Licenses

Scientific, Research, and Academic UseCommercial Use
Open Source License (AGPLv3) Includes reference software and associated intellectual property
No cost
Includes reference software and associated intellectual property Use internally – no cost
Distribute derivative software – no cost if released in accordance with the AGPLv3
Trial License Includes reference software and associated intellectual property without AGPLv3 requirements
No commercial rights
No cost
Includes reference software and associated intellectual property without AGPLv3 requirements
No commercial rights
No cost
Commercial IP License optionsWe agree not to assert our IP for non-commercial usesIncludes commercial rights to intellectual property and use of the reference software
Validation License – no cost, 12 month term
Simple License – quarterly fee, perpetual
Custom License – contact us

If you have any questions about our licenses, visit our FAQ page or contact us at licenses@numenta.com.

Partners

Numenta Licensing and Partners

Numenta works with partners to bring the power of HTM to the market. While we focus on the science and the core technology, we select application partners who have deep domain knowledge and are able to add an application layer tuned to market needs.

We are flexible in structuring these relationships in a way that works for both parties. If you are interested in becoming a partner, please review our License Guide , and email us at licenses@numenta.com.

Numenta Icon

Cortical.io

Cortical.io is leading the next generation of natural language processing: language intelligence. Founded on core principles of HTM, Cortical.io’s Semantic Folding technology translates text into sparse distributed representations. This enables a host of exciting applications that have challenged computer scientists for decades including sentiment analysis, automatic summarization, semantic search and conversational dialogue systems.

Grok

Grok is using HTM technology for advanced IT anomaly detection. Grok applies Numenta’s breakthrough technology to solving the IT department’s hardest problems, with a complete enterprise solution. Its modern user interface makes it easy to drill down to important anomalies and take action before a problem worsens.

Intelletic Trading Systems (ITS)

Intelletic Trading Systems (ITS) has developed an artificial intelligence platform using HTM for autonomous trading of futures and other financial assets. The company’s unique cortical learning approach is designed to generate greater profit and incur less risk than a human discretionary trader.

Close Section

Careers

We may be busy contemplating cortical theory, but we’ve got a work-hard, play-hard attitude. Our kitchen is stocked with snacks, and we enjoy weekly catered lunches from a variety of local restaurants.

At the heart of the peninsula, our downtown Redwood City location is a short walk from the Caltrain station. Outside the office, we blow off steam with company events, happy hours, and other fun activities. In the past, we’ve cheered on the SF Giants, baked pies at Pie Ranch in Pescadero, and do-si-doed through the night.

We currently have one full time position open. In addition we are always looking for strong research candidates to join us through our research internships. We also welcome young researchers, established professors, and scientists to join our Visiting Scholar Program.

Management Team

Numenta Door

Donna Dubinsky

CEO & Co-Founder

Donna is a serial entrepreneur best known for her work as CEO of Palm Computing and then Handspring, pioneers of the first successful handheld computers and smartphones. Previously, Donna spent 10 years in a multitude of sales, sales support, and logistics functions—both at Apple and at Claris, an Apple software subsidiary. She founded Numenta with her long-time business partner, Jeff Hawkins, in 2005.

Donna earned a B.A. from Yale University, and an M.B.A. from Harvard Business School. In addition to chairing Numenta’s board, she currently serves on the boards of Cortical.io (Vienna, Austria), Stanford Health Care (Palo Alto, CA), and Twilio (NYSE: TWLO). Donna also served on the board of Yale University from 2006-2018, including two years as Senior Fellow.

Donna Dubinsky

Jeff Hawkins

Co-Founder

Jeff is a scientist whose life-long interest in neuroscience led to the creation of Numenta and its focus on neocortical theory. His research focuses on how the cortex learns predictive models of the world through sensation and movement. In 2002, he founded the Redwood Neuroscience Institute, where he served as Director for three years. The institute is currently located at U.C. Berkeley. Previously, he co-founded two companies, Palm and Handspring, where he designed products such as the PalmPilot and Treo smartphone. In 2004 he wrote “On Intelligence”, a book about cortical theory.

Jeff earned his B.S. in Electrical Engineering from Cornell University in 1979. He was elected to the National Academy of Engineering in 2003.

Jeff Hawkins

Subutai Ahmad

VP of Research

Subutai brings experience across real time systems, computer vision and learning to Numenta. He has previously served as VP Engineering at YesVideo, Inc. where he helped grow the company from a three-person start-up to a leader in automated digital media authoring. In 1997, Subutai co- founded ePlanet Interactive, a spin-off from Interval Research. ePlanet developed the IntelPlay Me2Cam, the first computer vision product developed for consumers. He has served as a key researcher at Interval Research.

Subutai holds a B.S. in Computer Science from Cornell University, and a Ph.D in Computer Science from the University of Illinois at Urbana-Champaign. While pursuing his Ph.D, Subutai completed a thesis on computational neuroscience models of visual attention.

Subutai Ahmad

Christy Maver

VP of Marketing

Christy brings nearly two decades of technology marketing and communications experience to Numenta. Previously, she launched analytics programs for the Retail and Healthcare industries as the Global Product Marketing Director of Analytics at Actian. Christy held a number of software marketing roles during her 13 years at IBM, where she managed user groups, produced live demos and developed big data video tutorials. She was also one of the founding members of IBM’s thought leadership group: the IBM Institute for Business Value.

Christy holds a BA in Economics from Princeton University.

Christy Maver

Board of Directors

Ed Colligan

Former President & CEO, Palm, Inc.

Ed has been a part of the core team of five Silicon Valley start-ups. Ed’s first big success was Radius, Inc. where he was instrumental in building products and the brand. After Radius, Ed was the first vice president of marketing for Palm where he helped develop the original Palm Pilot, the Palm brand and Palm’s product strategy. He moved on from Palm to found Handspring where Ed and his partners created the forbearer of all future smartphones; the Handspring Treo. Ed drove the transaction that reunited Palm and Handspring into a single Palm again. He established relationships with wireless carriers globally and drove Palm’s annual smartphone business to more than $2 billion. As the CEO of Palm, Ed spearheaded the transformation that created the WebOS platform and Palm Pre line of smartphones.

Ed now spends his time investing in and mentoring entrepreneurs. Ed is a board member of Numenta, Inc., Active Mind Technology, and POPS Worldwide, and is an investor and on the board of advisors of six other start-up companies. Ed holds a B.S. from the University of Oregon.

Ed Colligan

Donna Dubinsky

CEO & Co-Founder

Donna is a serial entrepreneur best known for her work as CEO of Palm Computing and then Handspring, pioneers of the first successful handheld computers and smartphones. Previously, Donna spent 10 years in a multitude of sales, sales support, and logistics functions—both at Apple and at Claris, an Apple software subsidiary. She founded Numenta with her long-time business partner, Jeff Hawkins, in 2005.

Donna earned a B.A. from Yale University, and an M.B.A. from Harvard Business School. In addition to chairing Numenta’s board, she currently serves on the boards of Twilio (NYSE: TWLO) and Cortical.io (Vienna, Austria). Donna also served on the board of Yale University from 2006-2018, including two years as Senior Fellow.

Donna Dubinsky

Mike Farmwald

General Partner, Skymoon Ventures

Mike is a successful serial entrepreneur. He has founded many companies with breakthrough technologies including FTL, a super computing company that merged with MIPs, Epigram, which was acquired by Broadcom, Rambus and Matrix Semiconductor, a creator of 3D integrated circuits.

Mike currently sits on the board of Rambus (NASDAQ: RMBS). He is participating on the Numenta board as an individual, rather than as a representative of Skymoon Ventures. Mike holds a B.S. degree in Mathematics from Purdue University and a Ph.D. in Computer Science from Stanford University.

Mike Farmwald

Jeff Hawkins

Co-Founder

Jeff is a scientist whose life-long interest in neuroscience led to the creation of Numenta and its focus on neocortical theory. His research focuses on how the cortex learns predictive models of the world through sensation and movement. In 2002, he founded the Redwood Neuroscience Institute, where he served as Director for three years. The institute is currently located at U.C. Berkeley. Previously, he co-founded two companies, Palm and Handspring, where he designed products such as the PalmPilot and Treo smartphone. In 2004 he wrote “On Intelligence”, a book about cortical theory.

Jeff earned his B.S. in Electrical Engineering from Cornell University in 1979. He was elected to the National Academy of Engineering in 2003.

Jeff Hawkins

Harry Saal

Chairman, Retrotope, Inc.

In 2002, Dr. Harry J. Saal was chosen by the US Department of Justice to lead the Technical Committee charged with monitoring and enforcing the Microsoft Antitrust case, and he served as Chairman of the Committee through the May 2011 expiration of the Judgment.

Harry founded Nestar Systems, a pioneer in local area network systems, in 1978. In 1986, Harry became the founder and CEO of Network General Corporation, the first company wholly dedicated to the area of network diagnostics. From 1993 through 1995, Harry served as founding CEO and President of Smart Valley, Inc., a non-profit organization chartered to create a regional electronic community based on an advanced information infrastructure and the collective ability to use it.

Harry is active in philanthropy and community affairs, and serves on the board of the American Institute of Mathematics, among others. Harry holds a B.A., M.A. and Ph.D. in Physics from Columbia University.

Harry Saal

Close Section

Want to get in contact with us? Send us a message by filling out this form or use the appropriate email.

Fill out my online form.

Press Contact:
press@numenta.com

Licensing Inquiries:
licenses@numenta.com

Office

Phone 650-369-8282
Fax 650-369-8283
Address 791 Middlefield Road
Redwood City, CA 94063
Map Google Maps Link
  • Reception is on the second floor on the left
  • Street parking is available on surrounding streets
  • The Jefferson Avenue Garage is the nearest parking structure

Follow

Map to Numenta

Close Section