When facing a chemistry problem with many outputs, you might want to optimise one output first, like the yield of a reaction, or look for trade-offs for several outputs in a multi-objective fashion. In the latter, objectives/outputs may not be equally important (even across iterations) and very often, a human ranks these outputs to help with future experimentation. Here we will use Bayesian optimization with preference exploration, BOPE [1,2] (implemented in Botorch), to guide optimisation in a synthetic multi-output dataset, and we will apply it later to a Fischer-Tropsch synthesis dataset [3,4]. This algorithm selects importance (or utility) of the outputs based on pairwise comparison, and the small twist is that this comparison will be done by a large language model (LLM) instead of a defined utility function [5]. We welcome anyone that wants to contribute or know more about preferential Bayesian optimization, multi-input multi-output datasets and LLM prompting.

Check out our social media submission posts on LinkedIn and X!

References:

1. BoTorch tutorial for BOPE.
2. Lin ZJ, Astudillo R, Frazier P, Bakshy E. Preference Exploration for Efficient Bayesian Optimization with Multiple Outcomes. In: Proceedings of The 25th International Conference on Artificial Intelligence and Statistics. PMLR. pp. 4235–4258.
3. Lozano-Blanco G, Thybaut JW, Surla K, Galtier P, Marin GB. Single-Event Microkinetic Model for Fischer−Tropsch Synthesis on Iron-Based Catalysts. Ind Eng Chem Res 2008;47:5879–5891.
4. Chakkingal A, Janssens P, Poissonnier J, Virginie M, Khodakov AY, et al. Multi-output machine learning models for kinetic data evaluation : A Fischer–Tropsch synthesis case study. Chemical Engineering Journal 2022;446:137186.
5. Qin Z, Jagerman R, Hui K, Zhuang H, Wu J, et al. Large Language Models are Effective Text Rankers with Pairwise Ranking Prompting. DOI: 10.48550/arXiv.2306.17563.