Back

Laminar RNNs: using biologically-inspired network topology on the cortical laminar level in memory tasks

Shamir, I.; Assaf, Y.

2024-05-08 neuroscience
10.1101/2024.05.06.592696 bioRxiv
Show abstract

Advancements in neuroscience and artificial intelligence have been fueling one another for decades. In this study, we integrate a neuroimaging model of laminar-level connectomics into a biologically-inspired deep learning model of recurrent neural networks (RNNs) for working memory tasks. The resulting model offers a way to incorporate a more comprehensive representation of brain topology into artificial intelligence without diminishing the performance of the network compared to previous models.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.