Introduction

Defining Network Structures

The connectivity matrices between different layers are core to the modeling of the brain. They define the synaptic strength, positivity, plasticity, and other properties. In nn4n, these specific properties are applied to the corresponding connectivity matrices through masks. For complex structures such as excitatory-inhibitory constrainted RNN, the masks of adjacent layers are intercorrelated. The nn4n.mask modules provide some implementations to generate the masks for these complex structures.

Excitatory-Inhibitory Constrainted RNN

Multi-Area RNN

Multi-Area EI RNN

List of Masks

• BaseStruct
• MultiArea
• MultiAreaEI
• RandomInput