GAN

• GAN-Drug-Generator: Proposes a framework based on Feedback Generative Adversarial Network (GAN) for the generation and optimization of drug-like molecules, including a multiobjective optimization selection technique.
  
• MoFlowGAN: MoFlowGAN is a normalizing flow for molecular graphs that is heuristically biased towards easily synthesized, drug-like molecules, aiming to generate high-quality molecular graphs through a process similar to GANs.
  
• MolFilterGAN: MolFilterGAN is a progressively augmented generative adversarial network for triaging AI-designed molecules, focusing on improving the quality of generated molecules by filtering out undesired candidates early in the generation process.
  
• mol-Zero-GAN: Aims at optimizing pretrained generative models for drug candidate generation using Bayesian optimization.
  
• RRCGAN: RRCGAN combines a generative adversarial network with a regressor to generate small molecules with targeted properties, emphasizing the use of deep learning models to design molecules with specific desired attributes.
  
• SpotGAN: SpotGAN, a PyTorch implementation of a reverse-transformer GAN, generates scaffold-constrained molecules with property optimization, demonstrating advanced capabilities in generating molecules that adhere to specific structural constraints while optimizing for desired properties.