Lex Fridman
2 HN recommenders · recommended in 4 months, 2019–2021
2:09:42 · 329K views · Jul 1, 2019
Lex Fridman interviews Jeff Hawkins, founder of Numenta and author of On Intelligence, about the Thousand Brains Theory and reverse-engineering the neocortex. They also cover Hierarchical Temporal Memory and brain-inspired AI architectures.
HN people send others here to hear Hawkins explain his theory in his own words, including reference frames and the old versus new brain.
4 comments from Hacker News
Highly recommend the Lex Fridman interview to listen to Jeff explain live: this video While Numenta doesn't have the amazing results of DeepMind and friends, I think they are doing really great stuff in the space of biologically plausible intelligence.
I wouldn't say that the Triune brain theory is a 'myth'. It's just not accurate, but it's not a complete lie either. There certainly are different parts of the brain that have evolved at different times, and have different functions. The older parts have evolved to specific functions, while neocortex, which is found in mammals, has a uniform structure and is capable of learning and adapting to different tasks. 75% of human brain is neocortex. Jeff Hawkings explains the difference between the 'old parts' and 'new parts' pretty well in this interview: this video
Has anyone (beyond maybe self-driving software) tried using object tagging as a way to start introducing physics into a scene? E.g. human and bicycle have same motion vector, increases likelihood that human is riding bicycle. Bicycle and human have size and weight ranges that could be used to plot trajectory. Bicycles riding in a straight line and trees both provide some cues as to the gravity vector in the scene. Etc. etc. Seems like the camera motion is probably already solved with optical flow/photogrammetry stuff, but you might be able to use that to help scale the scene and start filtering your tagging based on geometric likelihood. The idea of hierarchical reference frames (outlined a bit by Jeff Hawkins here this video ) seems pretty compelling to me for contextualizing scenes to gain comprehension. Particularly if you build a graph from those reference frames and situate models tuned to the type of object at the root of each each frame (vertex). You could use that to help each model learn, too. So if a bike model projects a 'riding' edge towards the 'person' model, there wouldn't likely be much learning. e.g. [Person]-(rides)->[Bike] would have likely been encountered already. However if the [Bike] projects the (rides) edge towards the [Capuchin] sitting in the seat, the [Capuchin] model might learn that capuchins can (ride) and furthermore they can (ride) a [Bike].
There's a youtube generated one if you go to this video and click the three dots under the vid. I don't know if there's a way to get those things into an easy to read format? Clicking select all and pasting to the text editor sort of works but could be better.