Reaction
A conceptual route to biologically plausible approximate gradient descent
A user says the mechanism needn't be hardcoded, though nothing like it has been shown to organize in real brains.
TLDR
A user finds a proposed mechanism for approximate stochastic gradient descent compelling, saying it suggests biological backpropagation could be possible without being hardcoded. They note that nothing like it has been demonstrated to organize in real brains.
Combined views
—
1 Source, first seen ago
— likes— comments— saves— reposts