Fig. 5: Experimental result of classification and generative AI tasks. | Nature Communications

Fig. 5: Experimental result of classification and generative AI tasks.

From: Seamless optical cloud computing across edge-metro network for generative AI

Fig. 5

a The accuracy of MNIST handwritten digital image classification with different precision. b Confusion matrix for MNIST handwritten digital image classification. c Architecture of optical cloud computing system adapted for various tasks. d Convolved waveforms from the first layer of the map edge detection task, with the red and blue lines representing the ideal and experimentally generated waveforms, respectively. e Performance comparison between simulation and experiment with 6 different generative AI tasks (edges2handbags, edges2portrait, edges2shoes, map2edges, pix2pix-depth and segmentation). f Image generation results for season transfer task (winter to summer; summer to winter).

Back to article page