1 paper · 1 filter
Mattia Scardecchia
Despite the striking successes of deep neural networks trained with gradient-based optimization, these methods differ fundamentally from their biological counterparts. This gap rai…