Please read this before posting on this forum.
Table of Contents
- What is This Forum?
- Before Posting
- Getting started
- Forum Preferences
- Moderation Policies
- Examples of High Quality Posts
- Questions or Feedback Regarding this Forum
You may have arrived here after:
- Reading “On Intelligence”
- Seeing a presentation by one of the Numenta staff
- Reading a reference to Numenta in the media
In any case, if you are someone who is curious about the human brain, and particularly how intelligence arises from its biology, you will find yourself in good company here. Please come and Introduce yourself!
The purpose of this welcome guide is to prepare you to make the most of your time learning more about HTM, and being part of this community.
This forum centers around the machine intelligence concept of Hierarchical Temporal Memory (HTM) developed by Numenta and incubated in Numenta’s open source community. HTM is a biologically constrained theory of intelligence based upon research into the neuroscience of the mammalian neocortex.
- Visit the TOP page to see the most actively discussed topics in the last year / month / week.
- Visit the CATEGORIES page to get a broad overview of what’s going on here.
- Search before opening a new topic, as someone might have already created a topic. (Use the search icon in the upper right of every page.
Here are some topics you might be interested in if you are new to the site, just pick the heading that best describes your reason for joining.
I’m here because I want to learn more about HTM theory
It would be a crime not to watch HTM School, it has been put together specifically to get people up to speed with the theory, and provides a good knowledge foundation to build from.
If you have a question about a particular aspect, search these forums. Most aspects of the theory have been picked apart at length over the years. But if you can’t find the answers you’re after, create a new post.
I’m here because I want to debate the merits of the theory or argue for another approach
You are certainly free to do so, however you will find that most people here are quite deliberately Biological Inspirationalists. This doesn’t mean we hate neural nets or don’t understand deep learning; in fact many of us study them, use them, and recommend them as current state of the art for most learning tasks. It’s just we believe that understanding how nature’s implementation of intelligence works is the best first step to building machines that can properly and efficiently generalize their learning.
I’m here with some data I’d like to feed through an HTM system to try it out
A great place to start cold is to try out HTM Studio - it’s designed to be a quick, friendly, visual way to use NuPIC to do anomaly detection.
If you know Python and you’re used to working in notebook environments, a member of the community has built a NuPIC notebook for beginners.
I’m here because I want to build something using HTM
That’s great news - the community needs builders!
First, a word of caution: if you’re not a proficient software developer, then trying out HTM is not a good way to learn basics of coding.
If you have trouble using NuPIC (or another implementation) but your problem is due to lack of general programming experience, you may find difficulty getting answers. This isn’t because we don’t want to help you, it’s just that we are a relatively small community focused on a research niche, so we don’t have the collective resources to help with learning to code.
It might be overkill to send beginners on a 4 year Computer Science degree, but you should be familiar with the following:
- Algorithms and data structures (example)
- Computer programming, e.g. branching, abstractions, recursion, debugging, OOP, inheritance. (example, using Python)
Conversely, if you’ve already achieved mastery in the digital domain but are getting lost in all the biology talk, don’t be put off. It’s worth taking the time to fill the gaps in your neuroscience knowledge until you can more easily follow the discussions - many HTM researchers started out this way.
You might not want to receive emails from this forum, or get notified about topics in some categories. You have a lot of control over this in your preferences, where you can mute categories and turn off emails. For questions about preferences, please see the #other-topics:site-feedback forum.
Most offenses will be dealt with initially with a warning. Subsequent offenses will result in a temporary account suspension. HTM Forum staff members have the authority to permanently ban any forum member for any reason without warning. See our complete Terms of Service for details.
Reasons we moderate posts
- Code of Conduct Violations: We will not tolerate any harassment in any form. All posts must meet the expectations established in our Code of Conduct.
- Off Topic Posts: Staff will move topics from one category into another if they are a better fit elsewhere. Staff may also split a topic into multiple topics if the conversation is going into two different directions. We have a lot of categories for discussion. If you think there should be more, please contact a staff member.
- Typos, formatting, broken links, etc.: Staff may edit posts to improve the formatting and links within the posts.
Reasons we might suspend your account
- Continued Off-Topic Posts: This is a forum about HTM. We want to keep all topics (except those specifically in the #other-topics area) related to HTM. If you continue to post off-topic posts after being warned, your account may be suspended.
- Rudeness: Staff will not tolerate rudeness in any form. We expect our members to be nice to each other. We are all human beings.
- Arguing With Staff: If you have an issue with our moderation policy, please contact a staff member directly. Arguing with staff in a public discussion will result in an account suspension.
Here are some examples of well-constructed topics:
- What does "the receptive field center" mean in SP?
- HTM Mini-Columns into Hexagonal Grids!
- Oscillatory “Thousand Brains” Mind's Eye For HTM?
- Basic Spatial Pooler Questions
Please submit any questions or feedback on the #other-topics:site-feedback section of the forum.