Figure 1
From: Adaptive nodes enrich nonlinear cooperative learning beyond traditional adaptation by links
From biological details to a schematic diagram representing the traditional learning by links (synapses) and the proposed learning by nodes. (a) A biological schema of an output neuron composed of a soma (gray circle, top) with two roots of dendritic trees (light-blue arrows), splitting into many dendritic branches (light-blue lines). The signal arriving from each of the connecting input neurons (gray circles, bottom) travels via its axon and its many branches (red lines) until terminating at the meeting points with the dendrites, the synapses (green stars), where some branches travel to other neurons. (b) A simplified schema of a where a dendritic tree and its branches are denoted by a horizontal light-blue line and synapses by green arrows. (c) A fully connected network composed of N = 7 nodes (gray circles) with 2 dendritic trees per node (light-blue). (c 1 ) Each node receives inputs from all other nodes via synapses (green stars). The number of adaptive parameters, synapses, is N*C, where C is the average input connectivity (O(N2) in dense networks). (c 2 ) Similar topology to c 1 , where the number of adaptive parameters is only O(N), the number of dendritic trees, N*ND, where ND is the average number of dendritic trees per node.
