Phase classification of multi-principal element alloys via interpretable machine learning

Abstract

There is intense interest in uncovering design rules that govern the formation of various structural phases as a function of chemical composition in multi-principal element alloys (MPEAs). In this paper, we develop a machine learning (ML) approach built on the foundations of ensemble learning, post hoc model interpretability of black-box models, and clustering analysis to establish a quantitative relationship between the chemical composition and experimentally observed phases of MPEAs. The originality of our work stems from performing instance-level (or local) variable attribution analysis of ML predictions based on the breakdown method, and then identifying similar instances based on k-means clustering analysis of the breakdown results. We also complement the breakdown analysis with Ceteris Paribus profiles that showcase how the model response changes as a function of a single variable, when the values of all other variables are fixed. Results from local model interpretability analysis uncover key insights into variables that govern the formation of each phase. Our developed approach is generic, model-agnostic, and valuable to explain the insights learned by the black-box models. An interactive web application is developed to facilitate model sharing and accelerate the design of MPEAs with targeted properties.

A Comparison of Explainable Artificial Intelligence Methods in the Phase Classification of Multi-Principal Element Alloys

“…While the importance of mean_MeltingT and mean_CovalentRadius variables are emphasized by the SHAP values, the BD contributions identify mean_NsValence and maxdiff_Electronegativity as important. According to our previous work 37 , the constituent elements of MPEA compositions that form in the FCC phase span the first and second rows of the d-block elements from the periodic table. These elements have a wide range of valence electron number.…”

“…There are papers in the published literature, where the BD method is used for post hoc model interpretability in a non-hierarchical learning setting 5,[34][35][36] . In this study, the BD results are discussed by referring to the BD data from our recent published work 37 , where the step-down method was used.…”

“…The MPEA phase classification problem is well-studied in the literature [39][40][41][42][43][44] , which provides us with sufficient domain knowledge to compare the outcomes from SHAP and BD results. Previously, we reported a ML study result about BD analysis based an ensemble of support vector machine (eSVM) along with the k-means clustering method, followed by CP analysis 37 . This work focuses on providing a detailed comparison between BD and SHAP from the perspective of local post hoc model interpretability of black-box models.…”

Emulating Quantum Dynamics with Neural Networks via Knowledge Distillation

“…They were also able to qualitatively show, that when system approaches the glass transition, the effective cut-off distance for the particle-particle interactions increases rapidly, a phenomena that is experimentally observed in glass systems. As an another example, Lee et al, in their work 53 , trained support vector machines to predict the preferred phase type for the multi-principal element alloys. Next, by interpreting the trained models, they were able to discover phase-specific relations between the chemical composition of the alloys and their experimentally observed phases.…”

Inferring topological transitions in pattern-forming processes with self-supervised learning

“…This limits the scope of this class of methods primarily to already know and at least partially explore cases. In contrast, unsupervised learning techniques [23][24][25][26][27][28][29][30] do not require hand labelling or other time consuming manual and potentially subjective interventions. The vast majority of these unsupervised approaches use dimensionality reduction and clustering to classify topological phases in latent space and detect the topological transitions.…”