Grand Challenge Questions
1) Are we close to understanding systems-level neural computation? Why or Why not?
2) What are the key scientific challenges or technologies for achieving an understanding of neural computation?
3) Are there applications for neuromimetic processing that can lead to better technologies immediately? What are they?
These questions are deliberately left vague and their interpretation is left entirely to each speaker. Our hope is that plenary speakers will use this small workshop as an opportunity to present more speculative notions about how truly Neuromimetic Information Processing and (ultimately) Synthetic Cognition might be achieved and what it will look like.
These are great questions, and I'm looking forward to the discussion. To get the ball rolling here are a few thoughts.
ReplyDeleteGrand Challenge Questions
1) Are we close to understanding systems-level neural computation?
I think that maybe yes.... (when I'm in my optimistic mode ;-))
Why or Why not?
Because the combination of STDP and temporally coded spiking information is a remarkable combination that seems to work beautifully - a point that I will try to illustrate.
2) What are the key scientific challenges or technologies for achieving an understanding of neural computation?
Fully understanding the potential of STDP and temporal coding for unsupervised learning architectures... ideally with a formal mathematical analysis (something which I am incapable of doing).
Understanding how brains can store information reliably over several decades without recall...
3) Are there applications for neuromimetic processing that can lead to better technologies immediately? What are they?
Possibly - memristor-based STDP implementations....
Simon
Let me open up question 1 with a strong statement to get things going. No, I believe we over-attribute to STDP and connection weights what may be attributable to dynamics. In other words using a carefully crafted generative model with fixed weights, connection-like changes can be demonstrated without any connection changes. This can be achieved even with tiny changes in the activation (dynamics) of interconnected neurons. We need to better understand the contribution of such dynamics to understand networks.
ReplyDelete