Neural-Network-Biased Genetic Algorithms for Materials Design: Evolutionary Algorithms That Learn

Abstract

Machine learning has the potential to dramatically accelerate high-throughput approaches to materials design, as demonstrated by successes in biomolecular design and hard materials design. However, in the search for new soft materials exhibiting properties and performance beyond those previously achieved, machine learning approaches are frequently limited by two shortcomings. First, because they are intrinsically interpolative, they are better suited to the optimization of properties within the known range of accessible behavior than to the discovery of new materials with extremal behavior. Second, they require large pre-existing data sets, which are frequently unavailable and prohibitively expensive to produce. Here we describe a new strategy, the neural-network-biased genetic algorithm (NBGA), for combining genetic algorithms, machine learning, and high-throughput computation or experiment to discover materials with extremal properties in the absence of pre-existing data. Within this strategy, predictions from a progressively constructed artificial neural network are employed to bias the evolution of a genetic algorithm, with fitness evaluations performed via direct simulation or experiment. In effect, this strategy gives the evolutionary algorithm the ability to "learn" and draw inferences from its experience to accelerate the evolutionary process. We test this algorithm against several standard optimization problems and polymer design problems and demonstrate that it matches and typically exceeds the efficiency and reproducibility of standard approaches including a direct-evaluation genetic algorithm and a neural-network-evaluated genetic algorithm. The success of this algorithm in a range of test problems indicates that the NBGA provides a robust strategy for employing informatics-accelerated high-throughput methods to accelerate materials design in the absence of pre-existing data.

From DFT to machine learning: recent approaches to materials science–a review

“…Crystal structure classification between brittle or ductile phases of intermetallic compounds with only atomic radii was also studied [393]. Patra et al described a new strategy called neural-network-biased genetic algorithm (NBGA) to accelerate the discovery of materials with desired properties [394]. It uses artificial neural networks to bias the evolution of a genetic algorithm using fitness evaluations performed via direct simulation or experiments.…”

Machine Learning-Based Experimental Design in Materials Science

“…Other available OED methods include evolutionary algorithms such as genetic algorithms [27,28]. Such methods are scalable, but they have many parameters to tune (such as crossover and mutation rates).…”

Genetic algorithms for computational materials discovery accelerated by machine learning

“…It is possible to limit the relaxation to maintain reliable fingerprints; this would require that the GA runs through more candidates as the steps taken on the PES will be smaller. Another possibility is to utilize ML schemes that can also take care of relaxation [37][38][39] ; some examples already exist for coupling these with a GA. 40,41…”