Will Greedy retweeted
Our #NeurIPS2022 paper was selected for an oral!
You can see a 5min teaser here:
shorturl.at/dKMSX
Updated version of our story after review here:
arxiv.org/abs/2206.11769
New preprint 🚨! We introduce a new model of deep learning-like (hierarchical) credit assignment in corticocortical networks. It captures more biological observations and is more learning efficient than previous models! 🧵
arxiv.org/abs/2206.11769
Will Greedy retweeted
Nice theory paper (by @somnirons ) suggesting that bursting may carry gradient information arxiv.org/pdf/2206.11769.pdf -- love the topic, that's what much of my PhD thesis was about.
Will Greedy retweeted
Single-phase deep learning in cortico-cortical networks
deepai.org/publication/singl…
by @WillGreedy et al.
#Backpropagation #NeuralNetwork
Will Greedy retweeted
I-011: Single-phase deep learning in cortico-cortical networks. We introduce a model of deep learning in exc-inh cortical networks (burstCCN) that does not require multi-phase learning, which is consistent with a range of experimental observations. @WillGreedy @HengWZhu





