IBM's Context Aware Learning

Hey all. Today at the NICE workshop, Campbell Scott gave a presentation on IBM’s Context Aware Learning. My take on it is they took the HTM theory as described in the 2011 whitepaper and actually built a version of it with a working hierarchy. I’m sure some of you have wondered about what such a network will do. I will link the video once it’s posted online, for now the only thing I can link is the slides used in their last year’s presentation:

Which are a bit outdated (especially the current state ones).


Thanks Sergey. We used to have a working relationship with the Dr. Wilke’s team at IBM, but they don’t keep in contact much with us anymore because of corporate blowback from this blog post where we called Watson “classic AI”. They continue to work with HTM, but they don’t call it that anymore. :man_shrugging:


When you have a clear HTM win in hardware it seems like shooting yourself in the foot to distance yourself from a possible great example of success.

Maybe it’s time to suck it up and make nice?


May I suggest “basic” as an alternative to “classic” for future blog posts?


I would expect this type of behavior from the various lovely denizens of the internet (such egos are really fun to watch), but I’m quite surprised by this shedding of thick skin on IBM’s part. Perhaps I am missing their point of view since they’ve done much to modernize their work, but the Watson that’s famous to larger public as a notable achievement was literally classic AI… Regardless the presentation looks awesome and I’m impressed by their recent hardware success.

Thanks for the link Sergey!

1 Like