Learning Latent Space Energy-Based Prior Model for Molecule Generation
This addresses molecule design for drug discovery, offering an implicit learning approach to avoid invalid samples, though it is incremental as it builds on prior generative models.
The authors tackled the problem of generating valid and unique molecules from SMILES strings by proposing a latent space energy-based prior model, achieving competitive validity and uniqueness with state-of-the-art models and nearly perfect distribution matching of structural and chemical features.
Deep generative models have recently been applied to molecule design. If the molecules are encoded in linear SMILES strings, modeling becomes convenient. However, models relying on string representations tend to generate invalid samples and duplicates. Prior work addressed these issues by building models on chemically-valid fragments or explicitly enforcing chemical rules in the generation process. We argue that an expressive model is sufficient to implicitly and automatically learn the complicated chemical rules from the data, even if molecules are encoded in simple character-level SMILES strings. We propose to learn latent space energy-based prior model with SMILES representation for molecule modeling. Our experiments show that our method is able to generate molecules with validity and uniqueness competitive with state-of-the-art models. Interestingly, generated molecules have structural and chemical features whose distributions almost perfectly match those of the real molecules.