Fig. 3 | Scientific Reports

Fig. 3

From: Inverse binary optimization of convolutional neural network in active learning efficiently designs nanophotonic structures

Fig. 3

The IBO characteristics depending on the learning rate (Lr). (a–d) The output of the error function as a function of the iteration of the IBO method at various (a) Lr = 0.25, (b) Lr = 0.5, (c) Lr = 1, and (d) Lr = 2. The surrogate function used for (a) to (d) is a pre-trained CNN at N = 24 and the 2,205 training dataset. The global minimum is obtained by the EE method.

Back to article page