Active learning-based exploration of the catalytic pyrolysis of plastic waste

Research in chemical engineering requires experiments, which are often expensive, time-consuming, and laborious. Design of experiments (DoE) aims to extract maximal information from a minimum number experiments. The combination DoE with machine learning leads the field active learning, results more flexible, multi-dimensional selection Active has not yet been applied reaction modeling, as most techniques still require an excessive amount data. In this work, novel framework called GandALF that combines Gaussian processes clustering is proposed validated for yield prediction. performance compared other strategies virtual case study hydrocracking. Compared these methods, outperforms state-of-the-art achieves 33%-reduction approach first also perform well data-scarce applications, demonstrated by selecting investigate ex-situ catalytic pyrolysis plastic waste. Both common DoE-technique, our methodology selected 18 effect temperature, space time, catalyst on olefin LDPE. were significantly informative than regular proving applicability modeling experimental campaigns.