Thank you for the heads-up about Jeffs talk. I managed to catch it, and it was absolutely fantastic! The insights he shared were truly inspiring, and I’m excited about the new open source project. It’s great to see such innovation and forward-thinking ideas being presented. Looking forward to more updates and discussions around this. So glad to have you here keeping us informed!
I don’t know where Numenta is going with this forum.
At this time I mostly monitor for flame wars and ban-hammer spam.
I am not a Numenta employee and my total compensation to this point is the highly coveted Numenta coffee cup.
5 Likes
@vclay Thanks for the honest response.
I have so many questions. What comes to mind first are;
There are great discoveries in TBT that I am grateful, but many more that I am not able to emphasize with justice, so apologies for this.
Was there any breakthrough in the TBP that is expected at least in theory to allow building practical and intelligent machines? Something that will be later on par with backpropagation-based algorithms? Was there any practical PoC that was using reference frames successfuly?
2 Likes
@Bitking, we truly appreciate your dedication to the HTM forum and don’t want you to think we are taking your efforts for granted. Please reach out to me directly (vclay@numenta.com) so we can figure out a good solution together.
@Jose_Cueto Those are some good questions. We have been implementing many ideas from the thousand brains theory (including reference frames) over the past 2 years and got some promising results. Based on our experiments so far, we think that this approach really has a shot and would be much more sample-efficient and robust than existing approaches (for sensorimotor tasks). Of course, there is still a lot to do and we hope to validate our approach in many real-world applications over the next years. We will release our code and progress soon so then you can convince yourself and let us know your thoughts 
2 Likes