Data-Efficient Molecular Generation with Hierarchical Textual Inversion

arXiv:2405.02845v1 Announce Type: new
Abstract: Developing an effective molecular generation framework even with a limited number of molecules is often important for its practical deployment, e.g., drug discovery, since acquiring task-related molecular data requires expensive and time-consuming experimental costs. To tackle this issue, we introduce Hierarchical textual Inversion for Molecular generation (HI-Mol), a novel data-efficient molecular generation method. HI-Mol is inspired by the importance of hierarchical information, e.g., both coarse- and fine-grained features, in understanding the molecule distribution. We propose …

abstract arxiv costs cs.lg data deployment discovery drug discovery experimental framework hierarchical issue molecules practical q-bio.mn textual type