Fig. 6
From: Enhancing network modularity to mitigate catastrophic forgetting
Learning with the proposed approach (Dog) a Categorical accuracy against the goal. Blue indicates the FG of Dog. The orange line indicates \(\hbox {interMOD}_{1.0}\). The purple line indicates \(\hbox {interMOD}_{0.5}\). The green line indicates \(\hbox {interMOD}_{0}\). b, c Categorical accuracy against the unlearned goals Cat and Ship. The weights of the initial neural network are from the last epoch of (a). e, d Pearson’s correlation coefficient (cc) against before and after learning the unlearned goals Cat and Ship. The cc values of the filter elements of the initial neural network and the last neural network in (b, c) are shown. Blue indicates the cc values of all the filter elements. Red indicates the cc values of elements in the filter corresponding to intra-module links. Purple indicates the cc values of elements in the filter corresponding to inter-module links