Uncertainty-aware phase fraction prediction and active-learning-guided out-of-domain discovery of refractory multi-principal element alloys

A. K. Shargh; C. D. Stiles; J. A. El-Awady

arxiv: 2604.18322 · v1 · submitted 2026-04-20 · ❄️ cond-mat.mtrl-sci

Uncertainty-aware phase fraction prediction and active-learning-guided out-of-domain discovery of refractory multi-principal element alloys

A. K. Shargh , C. D. Stiles , J. A. El-Awady This is my paper

Pith reviewed 2026-05-10 03:58 UTC · model grok-4.3

classification ❄️ cond-mat.mtrl-sci

keywords refractory multi-principal element alloysphase fraction predictionmixture density networksactive learninguncertainty quantificationCALPHADmaterials discoveryout-of-domain prediction

0 comments

The pith

Mixture density networks predict phase fractions in refractory alloys while quantifying uncertainty to guide active learning toward compositions with unseen elements.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops a machine learning method using mixture density networks to forecast the fractions of different phases in refractory multi-principal element alloys across a wide temperature range. Separate models are trained for up to six phases on CALPHAD-derived data to capture the probabilistic nature of phase formation and report aleatoric uncertainty. A perturbation analysis identifies a minimal set of input features that preserves both accuracy and uncertainty calibration. The same uncertainty estimates then drive an active learning loop that proposes new alloy compositions containing elements absent from the training data, balancing exploration and exploitation to find ones with a desired target phase.

Core claim

We present a deep learning framework based on Mixture Density Networks to predict phase fractions in RMPEAs and quantify the associated aleatoric uncertainty across a wide temperature range. By training separate models for up to six constituent phases using CALPHAD derived data, our approach achieves high predictive accuracy while capturing the probabilistic nature of phase formation. To address epistemic uncertainty arising from incomplete knowledge of the most informative features, we perform a perturbation-based feature importance analysis and identify a minimally sufficient input set that maintains both predictive performance and uncertainty calibration. Finally, we propose an uncertainy

What carries the argument

Mixture Density Networks that output full probability distributions over phase fractions to capture aleatoric uncertainty, combined with a perturbation-based feature selection step and an uncertainty-driven active learning loop for out-of-domain composition suggestion.

If this is right

The reduced feature set identified by perturbation analysis sustains both high accuracy and reliable uncertainty estimates across temperatures.
Uncertainty-based active learning can surface alloy compositions outside the original training distribution that still form the target phase.
Separate per-phase models allow independent uncertainty tracking for each constituent phase in a multi-phase alloy.
The exploration-exploitation balance in the active learning strategy can be tuned to prioritize either novel or high-confidence candidates.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

Pairing the uncertainty estimates with a closed experimental loop could iteratively improve calibration by feeding real measurements back into the training set.
The same per-phase uncertainty framework might be applied to predict secondary properties such as hardness or oxidation resistance in the same alloy family.
If the active learning suggestions prove experimentally valid, the method offers a way to expand alloy design spaces without exhaustive enumeration of all possible multi-element combinations.

Load-bearing premise

CALPHAD-derived phase fraction data accurately reflects real experimental behavior and the model's uncertainty estimates remain well-calibrated for compositions that include elements never seen during training.

What would settle it

Laboratory measurements on active-learning-proposed alloys that contain previously unseen elements show measured phase fractions falling consistently outside the uncertainty intervals reported by the models.

read the original abstract

Refractory multi-principal element alloys (RMPEAs) represent a novel class of alloys characterized by an extensive compositional design space and the potential for exceptional mechanical performance under extreme conditions. While accurate phase stability prediction is essential for their robust design, existing machine learning approaches rely on deterministic mappings from composition-derived features to phase labels, neglecting the uncertainty inherent in such predictions. In this study, we present a deep learning framework based on Mixture Density Networks (MDNs) to predict phase fractions in RMPEAs and quantify the associated aleatoric uncertainty across a wide temperature range. By training separate models for up to six constituent phases of RMPEAs using CALPHAD derived data, our approach achieves high predictive accuracy while capturing the probabilistic nature of phase formation. To address epistemic uncertainty arising from incomplete knowledge of the most informative features, we perform a perturbation-based feature importance analysis and identify a minimally sufficient input set that maintains both predictive performance and uncertainty calibration. Finally, we propose an uncertainty-based active learning strategy to discover novel RMPEAs with the target phase incorporating previously unseen elements, while investigating the exploration-exploitation trade-off in model-guided discovery. Our uncertainty-aware framework has the potential to accelerate and improve the reliability of discovering novel high-performance alloys and is broadly applicable.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper applies separate MDNs per phase plus uncertainty-driven active learning to RMPEA phase fractions and out-of-domain discovery, but the extrapolation calibration is unproven.

read the letter

The paper trains separate Mixture Density Networks on CALPHAD data to predict fractions for up to six phases in refractory multi-principal element alloys, adds perturbation-based feature reduction, and runs an uncertainty-guided active learning loop to suggest compositions with elements absent from the training set. This is a straightforward pipeline that targets the practical need for screening large compositional spaces in high-temperature alloys. The per-phase modeling and explicit handling of aleatoric uncertainty fit the multi-phase nature of the problem, and the active learning component directly tackles exploration of unseen elements. The feature perturbation step is a reasonable way to keep inputs minimal without losing too much performance. The soft spots sit in the validation and the central extrapolation claim. The abstract asserts high accuracy and well-captured uncertainty but supplies no numbers, splits, baselines, or calibration plots, so those assertions cannot be checked. The stress-test concern is on target: MDNs produce aleatoric uncertainty by design, yet when the model encounters new elements the epistemic component dominates and the reported variances are unlikely to track actual errors. Without out-of-domain calibration checks or error-versus-uncertainty plots on held-out elements, the active learning loop risks prioritizing artifacts rather than real target-phase alloys. The reliance on CALPHAD as training source is standard but adds another layer of modeled data between the ML output and experiment. This work is aimed at computational materials scientists who screen refractory high-entropy alloys before lab work. A reader who wants a concrete example of MDN plus active learning in alloy design will get usable pipeline details even if the results need tightening. It deserves a serious referee because the domain application is relevant and the methods are applied coherently, even though revisions will be required for quantitative evidence on calibration and discovery performance. I would send it to review with specific requests for those checks.

Referee Report

2 major / 2 minor

Summary. The manuscript proposes a Mixture Density Network (MDN) framework trained on CALPHAD-derived data to predict phase fractions for up to six constituent phases in refractory multi-principal element alloys (RMPEAs), quantify aleatoric uncertainty across temperatures, apply perturbation-based feature selection to identify a minimal input set, and deploy an uncertainty-guided active learning loop to discover novel RMPEAs containing previously unseen elements.

Significance. If the reported predictive accuracy holds and the MDN uncertainties remain calibrated under extrapolation to new elements, the work could meaningfully accelerate alloy discovery by replacing deterministic classifiers with probabilistic models that also support targeted exploration-exploitation in large compositional spaces. The per-phase modeling and explicit handling of feature sufficiency are constructive steps beyond standard supervised learning in materials informatics.

major comments (2)

[Abstract] Abstract: the assertion that the MDN models 'achieve high predictive accuracy while capturing the probabilistic nature of phase formation' is presented without any quantitative metrics (MAE, R², log-likelihood, or calibration scores), baseline comparisons, or validation-split details. This directly undermines evaluation of the central performance claim.
[Active learning strategy] Active-learning strategy: the claim that uncertainty-based selection will discover valid target-phase RMPEAs with unseen elements assumes MDN aleatoric uncertainties remain well-calibrated under extrapolation. Because MDNs do not model epistemic uncertainty and CALPHAD coverage for refractory systems with novel elements is sparse, miscalibration could cause the loop to favor artifacts rather than true discoveries; explicit OOD calibration diagnostics (e.g., error-vs-uncertainty plots on held-out compositions containing new elements) are required to support this load-bearing step.

minor comments (2)

[Abstract] The abstract would be strengthened by stating the exact temperature range, the precise number of phases modeled, and at least one key performance number.
Notation for the MDN output parameters (means, variances, mixture weights) should be defined explicitly when first introduced to avoid ambiguity when discussing uncertainty propagation into the active-learning acquisition function.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and insightful comments, which have helped us improve the clarity and rigor of our manuscript. We provide point-by-point responses below and have revised the manuscript where appropriate to address the concerns.

read point-by-point responses

Referee: [Abstract] Abstract: the assertion that the MDN models 'achieve high predictive accuracy while capturing the probabilistic nature of phase formation' is presented without any quantitative metrics (MAE, R², log-likelihood, or calibration scores), baseline comparisons, or validation-split details. This directly undermines evaluation of the central performance claim.

Authors: We agree that the abstract should provide quantitative support for the performance claims to allow readers to evaluate them immediately. In the revised manuscript, we have updated the abstract to include key metrics such as MAE, R², and log-likelihood values from our validation procedure, along with a brief reference to the train-validation split and baseline comparisons. These details are expanded in the results section of the main text. revision: yes
Referee: [Active learning strategy] Active-learning strategy: the claim that uncertainty-based selection will discover valid target-phase RMPEAs with unseen elements assumes MDN aleatoric uncertainties remain well-calibrated under extrapolation. Because MDNs do not model epistemic uncertainty and CALPHAD coverage for refractory systems with novel elements is sparse, miscalibration could cause the loop to favor artifacts rather than true discoveries; explicit OOD calibration diagnostics (e.g., error-vs-uncertainty plots on held-out compositions containing new elements) are required to support this load-bearing step.

Authors: We appreciate the referee's emphasis on the need for OOD calibration diagnostics, as this is central to the reliability of the active learning loop. Our framework focuses on aleatoric uncertainty via MDNs, which is well-suited to the stochastic nature of phase formation, while the perturbation-based feature selection addresses aspects of epistemic uncertainty by identifying a minimal sufficient feature set. In the revised manuscript, we have added error-versus-uncertainty plots specifically for held-out compositions containing previously unseen elements to demonstrate calibration under extrapolation. We also include a discussion of the limitations of not modeling epistemic uncertainty explicitly (e.g., via ensemble or Bayesian methods) and note this as an avenue for future work. These additions support the active learning claims without overstatement. revision: partial

Circularity Check

0 steps flagged

No significant circularity; derivation relies on external CALPHAD data and standard MDN procedures

full rationale

The paper trains separate MDNs on CALPHAD-derived phase fraction data (external to the paper) to predict fractions and aleatoric uncertainty for up to six phases in RMPEAs. Perturbation-based feature selection identifies a minimal input set, and uncertainty is then used in an active-learning loop to propose out-of-domain compositions with unseen elements. No equations or steps reduce the reported predictions, uncertainties, or discovery proposals to quantities defined by parameters fitted inside the paper itself. The method applies off-the-shelf MDN and active-learning techniques to independent data without self-definitional loops, fitted-input renamings, or load-bearing self-citations. This is a standard, self-contained application.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Only the abstract is available, so the ledger is limited to the explicit data source and modeling assumptions stated there.

axioms (1)

domain assumption CALPHAD calculations supply reliable phase-fraction labels for training across the temperature range of interest
All models are trained on CALPHAD-derived data as stated in the abstract.

pith-pipeline@v0.9.0 · 5542 in / 1360 out tokens · 48574 ms · 2026-05-10T03:58:11.796836+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

56 extracted references · 56 canonical work pages

[1]

Zhang, T

Y. Zhang, T. T. Zuo, Z. Tang, M. C. Gao, K. A. Dahmen, P. K. Liaw, and Z. P. Lu, Microstructures and properties of high-entropy alloys, Prog. Mater. Sci. 61, 1 (2014)

work page 2014
[2]

T. Gao, J. Gao, S. Yang, and L. Zhang, Data-driven design of novel lightweight refractory high- entropy alloys with superb hardness and corrosion resistance, Npj Comput. Mater. 10, 256 (2024)

work page 2024
[3]

Čížek, O

J. Čížek, O. Melikhova, T. Vlasák, P. Hruška, D. Starỳ, and F. Lukáč, Characterization of lattice distortions in refractory metal complex concentrated alloys using positron annihilation spectroscopy, Materialia 23, 101450 (2022)

work page 2022
[4]

Miracle, High entropy alloys as a bold step forward in alloy development, Nat

D. Miracle, High entropy alloys as a bold step forward in alloy development, Nat. Commun. 10, 1805 (2019)

work page 2019
[5]

C. Wen, Y. Zhang, C. Wang, D. Xue, Y. Bai, S. Antonov, L. Dai, T. Lookman, and Y. Su, Machine learning assisted design of high entropy alloys with desired property, Acta Mater. 170, 109 (2019)

work page 2019
[6]

J. Wang, H. Kwon, H. S. Kim, and B.-J. Lee, A neural network model for high entropy alloy design, Npj Comput. Mater. 9, 60 (2023)

work page 2023
[7]

S. A. Giles, D. Sengupta, S. R. Broderick, and K. Rajan, Machine-learning-based intelligent framework for discovering refractory high-entropy alloys with improved high-temperature yield strength, Npj Comput. Mater. 8, 235 (2022)

work page 2022
[8]

Y. Ma, M. Li, Y. Mu, G. Wang, and W. Lu, Accelerated design for high-entropy alloys based on machine learning and multiobjective optimization, J. Chem. Inf. Model. 63, 6029 (2023)

work page 2023
[9]

A. K. Shargh and N. Abdolrahim, An interpretable deep learning approach for designing nanoporous silicon nitride membranes with tunable mechanical properties, Npj Comput. Mater. 9, 82 (2023)

work page 2023
[10]

H. Liu, A. K. Shargh, and N. Abdolrahim, Mining structure-property linkage in nanoporous materials using an interpretative deep learning approach, Materialia 21, 101275 (2022)

work page 2022
[11]

E. P. George, W. A. Curtin, and C. C. Tasan, High entropy alloys: A focused review of mechanical properties and deformation mechanisms, Acta Mater. 188, 435 (2020). 33

work page 2020
[12]

O. N. Senkov, G. B. Wilks, J. M. Scott, and D. B. Miracle, Mechanical properties of Nb25Mo25Ta25W25 and V20Nb20Mo20Ta20W20 refractory high entropy alloys, Intermetallics 19, 698 (2011)

work page 2011
[13]

H. Ren, R. Chen, X. Gao, T. Liu, G. Qin, S. Wu, and J. Guo, Development of wear-resistant dual- phase high-entropy alloys enhanced by C15 Laves phase, Mater. Charact. 200, 112879 (2023)

work page 2023
[14]

Y. Shi, B. Yang, and P. K. Liaw, Corrosion-resistant high-entropy alloys: a review, Metals 7, 43 (2017)

work page 2017
[15]

A. K. Shargh, C. D. Stiles, and J. A. El-Awady, Deep learning accelerated phase prediction of refractory multi-principal element alloys, Acta Mater. 283, 120558 (2025)

work page 2025
[16]

A. K. Shargh, C. D. Stiles, and J. A. El-Awady, Temperature-dependent discovery of BCC refractory multi-principal element alloys: Integrating deep learning and CALPHAD calculations, Comput. Mater. Sci. 259, 114186 (2025)

work page 2025
[17]

C. Chen, X. Han, Y. Zhang, P. K. Liaw, and J. Ren, Phase prediction of high-entropy alloys based on machine learning and an improved information fusion approach, Comput. Mater. Sci. 239, 112976 (2024)

work page 2024
[18]

Veeresham, N

M. Veeresham, N. Sake, U. Lee, and N. Park, Unraveling phase prediction in high entropy alloys: A synergy of machine learning, deep learning, and ThermoCalc, validation by experimental analysis, J. Mater. Res. Technol.-JMRT 29, 1744 (2024)

work page 2024
[19]

J. Qi, D. I. Hoyos, and S. J. Poon, Machine learning-based classification, interpretation, and prediction of high-entropy-alloy intermetallic phases, High Entropy Alloys Mater. 1, 312 (2023)

work page 2023
[20]

Z. He, H. Zhang, H. Cheng, M. Ge, T. Si, L. Che, K. Zheng, L. Zeng, and Q. Wang, Machine learning guided BCC or FCC phase prediction in high entropy alloys, J. Mater. Res. Technol. 29, 3477 (2024)

work page 2024
[21]

S. Wu, Z. Song, J. Wang, X. Niu, and H. Chen, Enhanced phase prediction of high-entropy alloys through machine learning and data augmentation, Phys. Chem. Chem. Phys. 27, 717 (2025)

work page 2025
[22]

G. Liu, Q. Wu, Y. Ma, J. Huang, Q. Xie, Q. Xiao, and T. Gao, Machine learning-based phase prediction in high-entropy alloys: further optimization of feature engineering, J. Mater. Sci. 60, 3999 (2025)

work page 2025
[23]

Mohammadi, J

A. Mohammadi, J. Tsang, X. Huang, and R. Kearsey, Convolutional neural network based methodology for flexible phase prediction of high entropy alloys, Can. Metall. Q. 64, 431 (2025)

work page 2025
[24]

C. M. Bishop, Mixture density networks, (1994)

work page 1994
[25]

D. C. Beaudry et al., Exceptional hardness in multiprincipal element alloys via hierarchical oxygen heterogeneities, Sci. Adv. 10, eado9697 (2024)

work page 2024
[26]

Tokarewicz and M

M. Tokarewicz and M. Grądzka-Dahlke, Review of recent research on AlCoCrFeNi high-entropy alloy, Metals 11, 1302 (2021)

work page 2021
[27]

X. Liu, J. Zhang, and Z. Pei, Machine learning for high-entropy alloys: Progress, challenges and opportunities, Prog. Mater. Sci. 131, 101018 (2023)

work page 2023
[28]

S. M. A. A. Alvi, M. Mulukutla, N. Flores, D. Khatamsaz, J. Janssen, D. Perez, D. Allaire, V. Attari, and R. Arroyave, Accurate and Uncertainty-Aware Multi-Task Prediction of HEA Properties Using Prior-Guided Deep Gaussian Processes, ArXiv Prepr. ArXiv250614828 (2025)

work page 2025
[29]

S. A. Giles, H. Shortt, P. K. Liaw, and D. Sengupta, Yield strength-plasticity trade-off and uncertainty quantification for machine-learning-based design of refractory high-entropy alloys, ArXiv Prepr. ArXiv230413932 (2023)

work page 2023
[30]

Zhang, X

J. Zhang, X. Liu, S. Bi, J. Yin, G. Zhang, and M. Eisenbach, Robust data-driven approach for predicting the configurational energy of high entropy alloys, Mater. Des. 185, 108247 (2020)

work page 2020
[31]

Novick, Q

A. Novick, Q. Nguyen, R. Garnett, E. Toberer, and V. Stevanović, Simulating high-entropy alloys at finite temperatures: An uncertainty-based approach, Phys. Rev. Mater. 7, 063801 (2023)

work page 2023
[32]

Y. Gu, C. D. Stiles, and J. A. El-Awady, A statistical perspective for predicting the strength of metals: Revisiting the Hall–Petch relationship using machine learning, Acta Mater. 266, 119631 (2024). 34

work page 2024
[33]

J. Luo, Y. Gu, Y. Wang, X. Ma, J. El-Awady, and others, Uncertainty-Aware Machine-Learning Framework for Predicting Dislocation Plasticity and Stress-Strain Response in FCC Alloys, ArXiv Prepr. ArXiv250620839 (2025)

work page 2025
[34]

M. Kim, M. Y. Ha, W.-B. Jung, J. Yoon, E. Shin, I. Kim, W. B. Lee, Y. Kim, and H. Jung, Searching for an optimal multi-metallic alloy catalyst by active learning combined with experiments, Adv. Mater. 34, 2108900 (2022)

work page 2022
[35]

C. K. Borg, E. S. Muckley, C. Nyby, J. E. Saal, L. Ward, A. Mehta, and B. Meredig, Quantifying the performance of machine learning models in materials discovery, Digit. Discov. 2, 327 (2023)

work page 2023
[36]

J. Hu, D. Liu, N. Fu, and R. Dong, Realistic material property prediction using domain adaptation based machine learning, Digit. Discov. 3, 300 (2024)

work page 2024
[37]

Pekala, E

M. Pekala, E. A. Pogue, K. McElroy, A. New, G. Bassen, B. Wilfong, J. Domenico, T. McQueen, and C. D. Stiles, Evaluating AI-Guided Design for Scientific Discovery, in AI for Accelerated Materials Design-NeurIPS 2023 Workshop (2023)

work page 2023
[38]

T. Yin, G. Panapitiya, E. D. Coda, and E. G. Saldanha, Evaluating uncertainty-based active learning for accelerating the generalization of molecular property prediction, J. Cheminformatics 15, 105 (2023)

work page 2023
[39]

E. A. Pogue et al., Closed-loop superconducting materials discovery, Npj Comput. Mater. 9, 181 (2023)

work page 2023
[40]

Wilfong et al., Ternary materials discovery using human-in-the-loop generative machine learning, RSC Adv

B. Wilfong et al., Ternary materials discovery using human-in-the-loop generative machine learning, RSC Adv. 15, 19126 (2025)

work page 2025
[41]

Andersson, T

J.-O. Andersson, T. Helander, L. Höglund, P. Shi, and B. Sundman, Thermo-Calc & DICTRA, computational tools for materials science, Calphad 26, 273 (2002)

work page 2002
[42]

Branco, L

P. Branco, L. Torgo, and R. P. Ribeiro, SMOGN: A Pre-Processing Approach for Imbalanced Regression, in First International Workshop on Learning with Imbalanced Domains: Theory and Applications (PMLR, 2017), pp. 36–50

work page 2017
[43]

J. Qi, A. M. Cheung, and S. J. Poon, High entropy alloys mined from binary phase diagrams, Sci. Rep. 9, 15501 (2019)

work page 2019
[44]

D. King, S. Middleburgh, A. McGregor, and M. Cortie, Predicting the formation and stability of single phase high-entropy alloys, Acta Mater. 104, 172 (2016)

work page 2016
[45]

Takeuchi and A

A. Takeuchi and A. Inoue, Classification of bulk metallic glasses by atomic size difference, heat of mixing and period of constituent elements and its application to characterization of the main alloying element, Mater. Trans. 46, 2817 (2005)

work page 2005
[46]

M. C. Troparevsky, J. R. Morris, P. R. Kent, A. R. Lupini, and G. M. Stocks, Criteria for predicting the formation of single-phase high-entropy alloys, Phys. Rev. X 5, 011041 (2015)

work page 2015
[47]

Vazquez, S

G. Vazquez, S. Chakravarty, R. Gurrola, and R. Arróyave, A deep neural network regressor for phase constitution estimation in the high entropy alloy system Al-Co-Cr-Fe-Mn-Nb-Ni, Npj Comput. Mater. 9, 68 (2023)

work page 2023
[48]

A. Y.-T. Wang, R. J. Murdock, S. K. Kauwe, A. O. Oliynyk, A. Gurlo, J. Brgoch, K. A. Persson, and T. D. Sparks, Machine learning for materials scientists: an introductory guide toward best practices, Chem. Mater. 32, 4954 (2020)

work page 2020
[49]

Shahriari, K

B. Shahriari, K. Swersky, Z. Wang, R. P. Adams, and N. De Freitas, Taking the human out of the loop: A review of Bayesian optimization, Proc. IEEE 104, 148 (2015)

work page 2015
[50]

Pelikan, D

M. Pelikan, D. Goldberg, and E. Cantu-Paz, BOA: the Bayesian optimization algorithm, vol I, (1999)

work page 1999
[51]

G.-J. Wang, C. Cheng, Y.-Z. Ma, and J.-Q. Xia, Likelihood-free inference with the mixture density network, Astrophys. J. Suppl. Ser. 262, 24 (2022)

work page 2022
[52]

D. P. Kingma, Adam: A method for stochastic optimization, ArXiv Prepr. ArXiv14126980 (2014)

work page 2014
[53]

Mandler and B

H. Mandler and B. Weigand, A review and benchmark of feature importance methods for neural networks, ACM Comput. Surv. 56, 1 (2024)

work page 2024
[54]

Zhang, P

Y. Zhang, P. Tiňo, A. Leonardis, and K. Tang, A survey on neural network interpretability, IEEE Trans. Emerg. Top. Comput. Intell. 5, 726 (2021). 35

work page 2021
[55]

S. A. Kube, C. Frey, C. McMullin, B. Neuman, K. M. Mullin, and T. M. Pollock, Navigating the BCC-B2 refractory alloy space: Stability and thermal processing with Ru-B2 precipitates, Acta Mater. 265, 119628 (2024)

work page 2024
[56]

Doucet, B

P. Doucet, B. Estermann, T. Aczel, and R. Wattenhofer, Bridging diversity and uncertainty in active learning with self-supervised pre-training, ArXiv Prepr. ArXiv240303728 (2024). 1 Supplementary Material for Uncertainty-aware phase fraction prediction and active-learning- guided out-of-domain discovery of refractory multi-principal element alloys Ali K. ...

work page 2024

[1] [1]

Zhang, T

Y. Zhang, T. T. Zuo, Z. Tang, M. C. Gao, K. A. Dahmen, P. K. Liaw, and Z. P. Lu, Microstructures and properties of high-entropy alloys, Prog. Mater. Sci. 61, 1 (2014)

work page 2014

[2] [2]

T. Gao, J. Gao, S. Yang, and L. Zhang, Data-driven design of novel lightweight refractory high- entropy alloys with superb hardness and corrosion resistance, Npj Comput. Mater. 10, 256 (2024)

work page 2024

[3] [3]

Čížek, O

J. Čížek, O. Melikhova, T. Vlasák, P. Hruška, D. Starỳ, and F. Lukáč, Characterization of lattice distortions in refractory metal complex concentrated alloys using positron annihilation spectroscopy, Materialia 23, 101450 (2022)

work page 2022

[4] [4]

Miracle, High entropy alloys as a bold step forward in alloy development, Nat

D. Miracle, High entropy alloys as a bold step forward in alloy development, Nat. Commun. 10, 1805 (2019)

work page 2019

[5] [5]

C. Wen, Y. Zhang, C. Wang, D. Xue, Y. Bai, S. Antonov, L. Dai, T. Lookman, and Y. Su, Machine learning assisted design of high entropy alloys with desired property, Acta Mater. 170, 109 (2019)

work page 2019

[6] [6]

J. Wang, H. Kwon, H. S. Kim, and B.-J. Lee, A neural network model for high entropy alloy design, Npj Comput. Mater. 9, 60 (2023)

work page 2023

[7] [7]

S. A. Giles, D. Sengupta, S. R. Broderick, and K. Rajan, Machine-learning-based intelligent framework for discovering refractory high-entropy alloys with improved high-temperature yield strength, Npj Comput. Mater. 8, 235 (2022)

work page 2022

[8] [8]

Y. Ma, M. Li, Y. Mu, G. Wang, and W. Lu, Accelerated design for high-entropy alloys based on machine learning and multiobjective optimization, J. Chem. Inf. Model. 63, 6029 (2023)

work page 2023

[9] [9]

A. K. Shargh and N. Abdolrahim, An interpretable deep learning approach for designing nanoporous silicon nitride membranes with tunable mechanical properties, Npj Comput. Mater. 9, 82 (2023)

work page 2023

[10] [10]

H. Liu, A. K. Shargh, and N. Abdolrahim, Mining structure-property linkage in nanoporous materials using an interpretative deep learning approach, Materialia 21, 101275 (2022)

work page 2022

[11] [11]

E. P. George, W. A. Curtin, and C. C. Tasan, High entropy alloys: A focused review of mechanical properties and deformation mechanisms, Acta Mater. 188, 435 (2020). 33

work page 2020

[12] [12]

O. N. Senkov, G. B. Wilks, J. M. Scott, and D. B. Miracle, Mechanical properties of Nb25Mo25Ta25W25 and V20Nb20Mo20Ta20W20 refractory high entropy alloys, Intermetallics 19, 698 (2011)

work page 2011

[13] [13]

H. Ren, R. Chen, X. Gao, T. Liu, G. Qin, S. Wu, and J. Guo, Development of wear-resistant dual- phase high-entropy alloys enhanced by C15 Laves phase, Mater. Charact. 200, 112879 (2023)

work page 2023

[14] [14]

Y. Shi, B. Yang, and P. K. Liaw, Corrosion-resistant high-entropy alloys: a review, Metals 7, 43 (2017)

work page 2017

[15] [15]

A. K. Shargh, C. D. Stiles, and J. A. El-Awady, Deep learning accelerated phase prediction of refractory multi-principal element alloys, Acta Mater. 283, 120558 (2025)

work page 2025

[16] [16]

A. K. Shargh, C. D. Stiles, and J. A. El-Awady, Temperature-dependent discovery of BCC refractory multi-principal element alloys: Integrating deep learning and CALPHAD calculations, Comput. Mater. Sci. 259, 114186 (2025)

work page 2025

[17] [17]

C. Chen, X. Han, Y. Zhang, P. K. Liaw, and J. Ren, Phase prediction of high-entropy alloys based on machine learning and an improved information fusion approach, Comput. Mater. Sci. 239, 112976 (2024)

work page 2024

[18] [18]

Veeresham, N

M. Veeresham, N. Sake, U. Lee, and N. Park, Unraveling phase prediction in high entropy alloys: A synergy of machine learning, deep learning, and ThermoCalc, validation by experimental analysis, J. Mater. Res. Technol.-JMRT 29, 1744 (2024)

work page 2024

[19] [19]

J. Qi, D. I. Hoyos, and S. J. Poon, Machine learning-based classification, interpretation, and prediction of high-entropy-alloy intermetallic phases, High Entropy Alloys Mater. 1, 312 (2023)

work page 2023

[20] [20]

Z. He, H. Zhang, H. Cheng, M. Ge, T. Si, L. Che, K. Zheng, L. Zeng, and Q. Wang, Machine learning guided BCC or FCC phase prediction in high entropy alloys, J. Mater. Res. Technol. 29, 3477 (2024)

work page 2024

[21] [21]

S. Wu, Z. Song, J. Wang, X. Niu, and H. Chen, Enhanced phase prediction of high-entropy alloys through machine learning and data augmentation, Phys. Chem. Chem. Phys. 27, 717 (2025)

work page 2025

[22] [22]

G. Liu, Q. Wu, Y. Ma, J. Huang, Q. Xie, Q. Xiao, and T. Gao, Machine learning-based phase prediction in high-entropy alloys: further optimization of feature engineering, J. Mater. Sci. 60, 3999 (2025)

work page 2025

[23] [23]

Mohammadi, J

A. Mohammadi, J. Tsang, X. Huang, and R. Kearsey, Convolutional neural network based methodology for flexible phase prediction of high entropy alloys, Can. Metall. Q. 64, 431 (2025)

work page 2025

[24] [24]

C. M. Bishop, Mixture density networks, (1994)

work page 1994

[25] [25]

D. C. Beaudry et al., Exceptional hardness in multiprincipal element alloys via hierarchical oxygen heterogeneities, Sci. Adv. 10, eado9697 (2024)

work page 2024

[26] [26]

Tokarewicz and M

M. Tokarewicz and M. Grądzka-Dahlke, Review of recent research on AlCoCrFeNi high-entropy alloy, Metals 11, 1302 (2021)

work page 2021

[27] [27]

X. Liu, J. Zhang, and Z. Pei, Machine learning for high-entropy alloys: Progress, challenges and opportunities, Prog. Mater. Sci. 131, 101018 (2023)

work page 2023

[28] [28]

S. M. A. A. Alvi, M. Mulukutla, N. Flores, D. Khatamsaz, J. Janssen, D. Perez, D. Allaire, V. Attari, and R. Arroyave, Accurate and Uncertainty-Aware Multi-Task Prediction of HEA Properties Using Prior-Guided Deep Gaussian Processes, ArXiv Prepr. ArXiv250614828 (2025)

work page 2025

[29] [29]

S. A. Giles, H. Shortt, P. K. Liaw, and D. Sengupta, Yield strength-plasticity trade-off and uncertainty quantification for machine-learning-based design of refractory high-entropy alloys, ArXiv Prepr. ArXiv230413932 (2023)

work page 2023

[30] [30]

Zhang, X

J. Zhang, X. Liu, S. Bi, J. Yin, G. Zhang, and M. Eisenbach, Robust data-driven approach for predicting the configurational energy of high entropy alloys, Mater. Des. 185, 108247 (2020)

work page 2020

[31] [31]

Novick, Q

A. Novick, Q. Nguyen, R. Garnett, E. Toberer, and V. Stevanović, Simulating high-entropy alloys at finite temperatures: An uncertainty-based approach, Phys. Rev. Mater. 7, 063801 (2023)

work page 2023

[32] [32]

Y. Gu, C. D. Stiles, and J. A. El-Awady, A statistical perspective for predicting the strength of metals: Revisiting the Hall–Petch relationship using machine learning, Acta Mater. 266, 119631 (2024). 34

work page 2024

[33] [33]

J. Luo, Y. Gu, Y. Wang, X. Ma, J. El-Awady, and others, Uncertainty-Aware Machine-Learning Framework for Predicting Dislocation Plasticity and Stress-Strain Response in FCC Alloys, ArXiv Prepr. ArXiv250620839 (2025)

work page 2025

[34] [34]

M. Kim, M. Y. Ha, W.-B. Jung, J. Yoon, E. Shin, I. Kim, W. B. Lee, Y. Kim, and H. Jung, Searching for an optimal multi-metallic alloy catalyst by active learning combined with experiments, Adv. Mater. 34, 2108900 (2022)

work page 2022

[35] [35]

C. K. Borg, E. S. Muckley, C. Nyby, J. E. Saal, L. Ward, A. Mehta, and B. Meredig, Quantifying the performance of machine learning models in materials discovery, Digit. Discov. 2, 327 (2023)

work page 2023

[36] [36]

J. Hu, D. Liu, N. Fu, and R. Dong, Realistic material property prediction using domain adaptation based machine learning, Digit. Discov. 3, 300 (2024)

work page 2024

[37] [37]

Pekala, E

M. Pekala, E. A. Pogue, K. McElroy, A. New, G. Bassen, B. Wilfong, J. Domenico, T. McQueen, and C. D. Stiles, Evaluating AI-Guided Design for Scientific Discovery, in AI for Accelerated Materials Design-NeurIPS 2023 Workshop (2023)

work page 2023

[38] [38]

T. Yin, G. Panapitiya, E. D. Coda, and E. G. Saldanha, Evaluating uncertainty-based active learning for accelerating the generalization of molecular property prediction, J. Cheminformatics 15, 105 (2023)

work page 2023

[39] [39]

E. A. Pogue et al., Closed-loop superconducting materials discovery, Npj Comput. Mater. 9, 181 (2023)

work page 2023

[40] [40]

Wilfong et al., Ternary materials discovery using human-in-the-loop generative machine learning, RSC Adv

B. Wilfong et al., Ternary materials discovery using human-in-the-loop generative machine learning, RSC Adv. 15, 19126 (2025)

work page 2025

[41] [41]

Andersson, T

J.-O. Andersson, T. Helander, L. Höglund, P. Shi, and B. Sundman, Thermo-Calc & DICTRA, computational tools for materials science, Calphad 26, 273 (2002)

work page 2002

[42] [42]

Branco, L

P. Branco, L. Torgo, and R. P. Ribeiro, SMOGN: A Pre-Processing Approach for Imbalanced Regression, in First International Workshop on Learning with Imbalanced Domains: Theory and Applications (PMLR, 2017), pp. 36–50

work page 2017

[43] [43]

J. Qi, A. M. Cheung, and S. J. Poon, High entropy alloys mined from binary phase diagrams, Sci. Rep. 9, 15501 (2019)

work page 2019

[44] [44]

D. King, S. Middleburgh, A. McGregor, and M. Cortie, Predicting the formation and stability of single phase high-entropy alloys, Acta Mater. 104, 172 (2016)

work page 2016

[45] [45]

Takeuchi and A

A. Takeuchi and A. Inoue, Classification of bulk metallic glasses by atomic size difference, heat of mixing and period of constituent elements and its application to characterization of the main alloying element, Mater. Trans. 46, 2817 (2005)

work page 2005

[46] [46]

M. C. Troparevsky, J. R. Morris, P. R. Kent, A. R. Lupini, and G. M. Stocks, Criteria for predicting the formation of single-phase high-entropy alloys, Phys. Rev. X 5, 011041 (2015)

work page 2015

[47] [47]

Vazquez, S

G. Vazquez, S. Chakravarty, R. Gurrola, and R. Arróyave, A deep neural network regressor for phase constitution estimation in the high entropy alloy system Al-Co-Cr-Fe-Mn-Nb-Ni, Npj Comput. Mater. 9, 68 (2023)

work page 2023

[48] [48]

A. Y.-T. Wang, R. J. Murdock, S. K. Kauwe, A. O. Oliynyk, A. Gurlo, J. Brgoch, K. A. Persson, and T. D. Sparks, Machine learning for materials scientists: an introductory guide toward best practices, Chem. Mater. 32, 4954 (2020)

work page 2020

[49] [49]

Shahriari, K

B. Shahriari, K. Swersky, Z. Wang, R. P. Adams, and N. De Freitas, Taking the human out of the loop: A review of Bayesian optimization, Proc. IEEE 104, 148 (2015)

work page 2015

[50] [50]

Pelikan, D

M. Pelikan, D. Goldberg, and E. Cantu-Paz, BOA: the Bayesian optimization algorithm, vol I, (1999)

work page 1999

[51] [51]

G.-J. Wang, C. Cheng, Y.-Z. Ma, and J.-Q. Xia, Likelihood-free inference with the mixture density network, Astrophys. J. Suppl. Ser. 262, 24 (2022)

work page 2022

[52] [52]

D. P. Kingma, Adam: A method for stochastic optimization, ArXiv Prepr. ArXiv14126980 (2014)

work page 2014

[53] [53]

Mandler and B

H. Mandler and B. Weigand, A review and benchmark of feature importance methods for neural networks, ACM Comput. Surv. 56, 1 (2024)

work page 2024

[54] [54]

Zhang, P

Y. Zhang, P. Tiňo, A. Leonardis, and K. Tang, A survey on neural network interpretability, IEEE Trans. Emerg. Top. Comput. Intell. 5, 726 (2021). 35

work page 2021

[55] [55]

S. A. Kube, C. Frey, C. McMullin, B. Neuman, K. M. Mullin, and T. M. Pollock, Navigating the BCC-B2 refractory alloy space: Stability and thermal processing with Ru-B2 precipitates, Acta Mater. 265, 119628 (2024)

work page 2024

[56] [56]

Doucet, B

P. Doucet, B. Estermann, T. Aczel, and R. Wattenhofer, Bridging diversity and uncertainty in active learning with self-supervised pre-training, ArXiv Prepr. ArXiv240303728 (2024). 1 Supplementary Material for Uncertainty-aware phase fraction prediction and active-learning- guided out-of-domain discovery of refractory multi-principal element alloys Ali K. ...

work page 2024