Covariance learning of correlated patterns in competitive networks
Neural Computation
How inhibitory oscillations can train neural networks and punish competitors
Neural Computation
Neural Networks - Special issue: Computational theories of the functions of the hippocampus
2005 Special issue: Interpreting hippocampal function as recoding and forecasting
Neural Networks - Special issue: Computational theories of the functions of the hippocampus
External activity and the freedom to recode
Neurocomputing
The hippocampus as a stable memory allocator for cortex
Neural Computation
| Hi-index | 0.00 |
We investigate the dynamics of a class of recurrent randomnetworks with sparse, asymmetric excitatory connectivity and globalshunting inhibition mediated by a single interneuron. Usingprobabilistic arguments and a hyperbolic tangent approximation tothe gaussian, we develop a simple method for setting the averagelevel of firing activity in these networks. We demonstrate throughsimulations that our technique works well and extends to networkswith more complicated inhibitory schemes. We are interestedprimarily in the CA3 region of the mammalian hippocampus, and therandom networks investigated here are seen as modeling the a prioridynamics of activity in this region. In the presence of externalstimuli, a suitable synaptic modification rule could shape thisdynamics to perform temporal information processing tasks such assequence completion and prediction.