|
|
|
|
|
|
by jerf
4200 days ago
|
|
|
If I'm following this presentation correct it seems like it would be relatively simple to add time into these models by using the previous frame's guesses as priors for this frame. He offhandedly mentioned something once that may have been this, but there wasn't enough context to be sure this is what he meant. Still, with the general idea of this reliable high-level affine invariance it doesn't seem hard to imagine how to convert this to a temporally-aware approach, at least at a basic level. |
|
|
Time is a tricky bastard. It is one thing to incorporate some Markovian dynamics in an artificial network, it is another thing to cope with dynamics as we know it. You might be interested in for example something like what Ralf Der is studying: http://www.informatik.uni-leipzig.de/~der/ (Tishby is working on this as well: https://www.cs.purdue.edu/homes/spa/venice08/docs/Venice-Tih... (pdf, slides)).
Two things I learned from his talk:
* If you have (human-like) part-based representations you need linear relationships to build up the whole again.
* Routing is key.
With dynamics involved, say movie scenes, or your hand moving in front of your own eyes, we might postulate that we will first discover linear relationships as well. A car disappearing temporarily behind a fence on the right we will expect to appear on the other side at the left.
Routing in the brain that corresponds with dynamic behavior, is probably something else than the time slicing / windowing we are all so familiar with in machine learning. It also goes beyond a simple central pattern generator for locomotion, which is not a routing problem at all (but just a clone of the outside frequency pattern with dynamics between neurons). "Real" dynamics is about being able to put items into slots, to learn a (visual) grammar. The thought that routing is key seems also be shared by Schmidthuber who created LTSM only to be able to route errors in a more sophisticated way through a network.