Rank dependency of rescaled pruning in recurrent neural networks
Alex Q. Wang, Soon Ho Kim, Hannah Choi
The paper shows that synaptic rescaling preserves low-dimensional dynamics in low-rank RNNs but fails in high-rank networks during pruning.
Author
Georgia Institute of TechnologyORCID 0000-0002-4769-3997Unclaimed ORCID profile · follow to get notified of new preprints
4 papers
Recent·Most cited
Alex Q. Wang, Soon Ho Kim, Hannah Choi
The paper shows that synaptic rescaling preserves low-dimensional dynamics in low-rank RNNs but fails in high-rank networks during pruning.
The paper develops a mean-field model to study how heterogeneous inhibitory cell types in V1 affect network dynamics and stability through spatial connectivity properties.
Yunmiao Wang, Soon Ho Kim, Ellie Jiayi He…·Dieter Jaeger
The paper uses voltage imaging to study fast cortical dynamics during a reaching task, identifying distinct oscillatory patterns and network behaviors.
Joseph Del Rosario, Stefano Coletta, Soon Ho Kim…·Bilal Haider
The paper investigates how lateral inhibition from PV and SST neurons in mouse V1 affects neural and perceptual contrast sensitivity, revealing distinct computational roles for each cell type.