Bayesian optimization with active learning of design constraints using an entropy-based approach

Abstract

The design of alloys for use in gas turbine engine blades is a complex task that involves balancing multiple objectives and constraints. Candidate alloys must be ductile at room temperature and retain their yield strength at high temperatures, as well as possess low density, high thermal conductivity, narrow solidification range, high solidus temperature, and a small linear thermal expansion coefficient. Traditional Integrated Computational Materials Engineering (ICME) methods are not sufficient for exploring combinatorially-vast alloy design spaces, optimizing for multiple objectives, nor ensuring that multiple constraints are met. In this work, we propose an approach for solving a constrained multi-objective materials design problem over a large composition space, specifically focusing on the Mo-Nb-Ti-V-W system as a representative Multi-Principal Element Alloy (MPEA) for potential use in next-generation gas turbine blades. Our approach is able to learn and adapt to unknown constraints in the design space, making decisions about the best course of action at each stage of the process. As a result, we identify 21 Pareto-optimal alloys that satisfy all constraints. Our proposed framework is significantly more efficient and faster than a brute force approach.

Evolution-guided Bayesian optimization for constrained multi-objective optimization in self-driving labs

“…Unknown constraints present a more challenging optimization scenario since the goal of efficiently achieving multiple objectives needs to be balanced with enough sampling of the feasibility boundaries to learn the constraints. This was previously demonstrated by Khatamsaz et al for alloy design [31], [37] as well as by Cao et al for liquid formulations [16], where in all cases an appropriate machine learning model needs to be trained a posteriori to classify whether samples are feasible or not. Based on the good performance of EGBO on both the self-driving AgNP platform and the synthetic problems, we now explore how to adapt the algorithmic framework to reduce sampling wastage.…”

Artificial Intelligence (AI) Workflow for Catalyst Design and Optimization

“…The EHVI function considers the entire Pareto front, providing a comprehensive view of all of the optimal solutions. It is similar to the improvement in the EI acquisition function used for single-objective optimization, the difference is that in EHVI, the improvement refers to the expected increase in the dominated hypervolume if a new sampling point were to be incorporated. , This means that EHVI considers not just the mean and variance predictions of the underlying Gaussian process model but also the correlation between multiple objectives. It guides exploration through the high-dimensional parameter space to find the next promising point for exploration.…”

“…In the Bayesian optimization framework applied to catalyst synthesis parameter optimization, the active learning loop is the crucial phase that bridges the gap between theoretical predictions and practical implementation. 44 Here, the selected set of parameters provided by the acquisition function is translated into a tangible experimental setup. Initially, experiments are conducted by using the suggested combination of parameters.…”

Navigating the Expansive Landscapes of Soft Materials: A User Guide for High-Throughput Workflows

“…Active sampling can more rapidly accommodate these restrictions and avoid the synthesis of a large library where a significant fraction may be unusable. Machine learning and Bayesian approaches are popular active learning schemes that have been successfully applied to diverse chemical challenges including the synthesis of metallic alloys , and nanoparticles, drug discovery, , catalyst development, and the evaluation of properties of bulk polymers ranging from electronic bandgap to thermal transitions . Examples of high-throughput synthesis coupled to iterative sampling include the design of protein-stabilizing random copolymers using automated polymer synthesis, , the identification of 19 F MRI contrast agents using continuous-flow chemistry, and the development of polymeric injectables for drug delivery …”