MANDREL: Modular Reinforcement Learning Pipelines for Material Discovery


  • Clyde Fare IBM Research Europe
  • George K. Holt STFC Hartree
  • Lamogha Chiazor IBM Research Europe
  • Michail Smyrnakis STFC Hartree
  • Robert Tracey IBM Research Europe
  • Lan Hoang IBM Research Europe



Artificial Intelligence, Software and testing tools for developing AI technologies, Simulation environments for AI agents and multi-agent systems


AI-driven materials discovery is evolving rapidly with new approaches and pipelines for experimentation and design. However, the pipelines are often designed in isolation. We introduce a modular reinforcement learning framework for inter-operable experimentation and design of tailored, novel molecular species. The framework unifies reinforcement learning (RL) pipelines and allows the mixing and matching of choices for the underlying chemical action space, molecular representation, desired molecular properties, and RL algorithm. Our demo showcases the framework's capabilities applied to benchmark problems like quantitative estimate of drug-likeness and PLogP, as well as the design of novel small molecule solvents for carbon capture.




How to Cite

Fare, C., Holt, G. K., Chiazor, L., Smyrnakis, M., Tracey, R., & Hoang, L. (2024). MANDREL: Modular Reinforcement Learning Pipelines for Material Discovery. Proceedings of the AAAI Conference on Artificial Intelligence, 38(21), 23787-23789.