One of the annoying things about my current research on temporality is that there are several big fat books everybody (correctly) regards as essential references that I don’t have the patience to read fully.
Conversation
Replying to
Depends on what you mean by that 🙂.
It is vague enough it could point to anything from specific computing architectures to Jeff Hawkins work to Proust.
1
3
The idea of temporality, or at least temporal memory, is lacking from today's deep networks. It's something that is baked into our brains' neural networks, and we have some ideas about how we construct models through sensory-motor interaction with reality.
1
2
3
In case you haven't read our latest work, it builds upon the HTM ideas of sequence memory.
1
1
3
Replying to
Ah yes, I have been meaning to catch up on latest from you guys. I think I saved that ref already but thanks for the re-up and reminder. It is definitely relevant.
Replying to
It's a fundamental change in the way we think hierarchy works in the brain. I'm working on a video about how the brain's hierarchy is different from deep learning hierarchies based on a global error propagation function. Out in one week on our YT channel
1
4

