In RNNs trained on memory tasks, abrupt learning is preceded by geometric restructuring of slow points, and a temporal consistency loss can trigger that restructuring and speed up training.
Attractor dynamics gate cortical information flow during decision-making
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
q-bio.NC 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
Understanding and controlling the geometry of memory organization in RNNs
In RNNs trained on memory tasks, abrupt learning is preceded by geometric restructuring of slow points, and a temporal consistency loss can trigger that restructuring and speed up training.