Skip to main content
Fig. 1 | Journal of Cheminformatics

Fig. 1

From: A de novo molecular generation method using latent vector based generative adversarial network

Fig. 1

Workflow of the LatentGAN. The latent vectors generated from the encoder part of the heteroencoder is used as the input for the GAN. Once the training of the GAN is finished, new compounds are generated by first sampling the generator network of the GAN and then converting the sampled latent vector into a molecular structure using the decoder component of the heteroencoder

Back to article page