Modular inhibitory coding in binary networks

We developed and characterized properties of new class of binary models where, as observed in biological networks, excitatory neurons are functionally separated from inhibitory units. New patterns, represented as activation and inactivation of binary units in excitatory layer, are stored in the network through recruitment and training of inhibitory units that are group into individual modules and interact with excitatory layer. We investigate the roles the two populations play in memory storage and show that inhibitory layer plays a critical role in memory storage and management and that capacity of this new type of network scales with number of inhibitory neurons. Further, we show that performance of the network is only gradually diminished when excitatory to excitatory connections are removed, but critically depends on inhibitory to excitatory connections. These results are in line with new experimental work showing that inhibitory interneurons are playing critical role in memory storage and recall in the brain networks and may also address why generally excitatory networks exhibit sparser reciprocal connectivity as compared to connections to/from inhibitory units. We further show advantages of so designed coding shame in terms of memory capacity, its expansion with progressive storage of new memories as well as network behavior for large memory loading.