You could talk about how the anomaly likelihood algorithm works and (briefly how it differs from other time series anomaly detection algorithms)?
It would be interesting to get an idea of a typical week at Numenta. Have a quick office walkthrough vid starting at the street, maybe stop at each desk and have each person give the three line summary of the project or question they’re working on at that moment. I don’t know if you have regular meetings, but a shot from the corner of the room during a meeting and a few lines of someone saying whatever is said at your meetings. HTM is amazing and I really love it. Numenta, as a functional collaboration of human brains, is also a really special project in itself.
I like both ideas from @Jonathan_Mackenzie and @Glenn_Gasner. I would also personally be curious to see an example of NuPIC being applied to a live data source. It seems like NuPIC (in the examples at least) is setup to pull records in one by one from a static csv file, though for a pure streaming situation the setup would need to be different right? I would definitely feel empowered by that as an aspiring HTM practitioner, though maybe it already exists somewhere?
I have examples of NuPIC processing live streams.
Awesome, yes I kinda figured they’d be there somewhere. Thanks!
I kindof don’t want to talk about anomaly detection because it is not biologically plausible.
This is also covered in quite a bit of detail in our most recent paper (open access):
I was unable to find the video on TM part 2 on your YouTube channel.
Could you please share the link if you have done it?
I really enjoyed the Interview with a Neuroscientist series, I’d be very happy to see more of those!
The next one will be “Interview With a Roboticist!”