Would be interesting to see if they are what they are used for
One of the success stories is Cortical IO.
They use some of the basic HTM concepts.
Does cortical.io use spatial pooling or temporal memory or just SDRs encoding based on semantic folding?
Cortical published a paper about how they build text retinas using corpuses of text. You’ll find the paper if you google it. They do not use SP/TM to build these retinas. That being said, the retinas they’ve created will be useful, I imagine, in many ways for the coming wave of biologically inspired intelligence.
Here’s the link to their paper about Semantic Folding:
Great paper, but they don’t explain how they organize their DB. There were some threats about that here and I think the conclusion was that they use SOM’s.