Yinsong Chen, Samson Yu, J. Eshraghian, Chee Peng Lim
Despite significant advances in deep neural networks across diverse domains, challenges persist in safety-critical contexts, including domain shift sensitivity and unreliable uncertainty estimation. To address these issues, this study investigates Bayesian learning for uncertainty handling in modern neural networks. However, the high-dimensional, non-convex nature of the posterior distribution poses practical limitations for epistemic uncertainty estimation. The Laplace approximation, as a cost-efficient Bayesian method, offers a practical solution by approximating the posterior as a multivariate normal distribution but faces computational bottlenecks in precise covariance matrix computation and storage. This research employs subnetwork inference, utilizing only a subset of the parameter space for Bayesian inference. In addition, a Kronecker-factored and low-rank representation is explored to reduce space complexity and computational costs. Several corrections are introduced to converge the approximated curvature to the exact Hessian matrix. Numerical results demonstrate the effectiveness and competitiveness of this method, whereas qualitative experiments highlight the impact of Hessian approximation granularity and parameter space utilization in Bayesian inference on mitigating overconfidence in predictions and obtaining high-quality uncertainty estimates.
{"title":"Sparse subnetwork inference for neural network epistemic uncertainty estimation with improved Hessian approximation","authors":"Yinsong Chen, Samson Yu, J. Eshraghian, Chee Peng Lim","doi":"10.1063/5.0193951","DOIUrl":"https://doi.org/10.1063/5.0193951","url":null,"abstract":"Despite significant advances in deep neural networks across diverse domains, challenges persist in safety-critical contexts, including domain shift sensitivity and unreliable uncertainty estimation. To address these issues, this study investigates Bayesian learning for uncertainty handling in modern neural networks. However, the high-dimensional, non-convex nature of the posterior distribution poses practical limitations for epistemic uncertainty estimation. The Laplace approximation, as a cost-efficient Bayesian method, offers a practical solution by approximating the posterior as a multivariate normal distribution but faces computational bottlenecks in precise covariance matrix computation and storage. This research employs subnetwork inference, utilizing only a subset of the parameter space for Bayesian inference. In addition, a Kronecker-factored and low-rank representation is explored to reduce space complexity and computational costs. Several corrections are introduced to converge the approximated curvature to the exact Hessian matrix. Numerical results demonstrate the effectiveness and competitiveness of this method, whereas qualitative experiments highlight the impact of Hessian approximation granularity and parameter space utilization in Bayesian inference on mitigating overconfidence in predictions and obtaining high-quality uncertainty estimates.","PeriodicalId":502250,"journal":{"name":"APL Machine Learning","volume":"35 12","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-04-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140740073","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Siqi Zhang, Zhoumo Zeng, Xiaocen Wang, Shili Chen, Yang Liu
The evaluation of well integrity in double-casing wells is critical for ensuring well stability, preventing oil and gas leaks, avoiding pollution, and ensuring safety throughout well development and production. However, the current predominant method of assessing cementing quality primarily focuses on single-casing wells, with limited work conducted on double-casing wells. This study introduces a novel approach for evaluating the cementing quality using the Inception module of convolutional neural networks. First, the finite-difference method is employed to generate borehole sonic data corresponding to a variety of model configurations, which are used to train a neural network that learns spatial features from the borehole sonic data to reconstruct the slowness model. By adjusting the network architecture and parameters, it is discovered that a neural network with two blocks and 4096 nodes in the fully connected layer demonstrated the best imaging results and exhibited strong anti-noise capabilities. The proposed method is validated using practical wellbore size models, demonstrating excellent results and offering a more effective means of evaluating wellbore integrity in double-casing wells. In addition, dipole acoustic logging data are used to conduct slowness model imaging of the compressional (P-) wave and shear (S-) wave in double-casing wells to verify the feasibility of cementing quality evaluation. The developed method contributes to more accurate evaluations of wellbore integrity for the oil and gas industry, leading to improved safety and environmental outcomes.
{"title":"Imaging in double-casing wells with convolutional neural network based on inception module","authors":"Siqi Zhang, Zhoumo Zeng, Xiaocen Wang, Shili Chen, Yang Liu","doi":"10.1063/5.0191452","DOIUrl":"https://doi.org/10.1063/5.0191452","url":null,"abstract":"The evaluation of well integrity in double-casing wells is critical for ensuring well stability, preventing oil and gas leaks, avoiding pollution, and ensuring safety throughout well development and production. However, the current predominant method of assessing cementing quality primarily focuses on single-casing wells, with limited work conducted on double-casing wells. This study introduces a novel approach for evaluating the cementing quality using the Inception module of convolutional neural networks. First, the finite-difference method is employed to generate borehole sonic data corresponding to a variety of model configurations, which are used to train a neural network that learns spatial features from the borehole sonic data to reconstruct the slowness model. By adjusting the network architecture and parameters, it is discovered that a neural network with two blocks and 4096 nodes in the fully connected layer demonstrated the best imaging results and exhibited strong anti-noise capabilities. The proposed method is validated using practical wellbore size models, demonstrating excellent results and offering a more effective means of evaluating wellbore integrity in double-casing wells. In addition, dipole acoustic logging data are used to conduct slowness model imaging of the compressional (P-) wave and shear (S-) wave in double-casing wells to verify the feasibility of cementing quality evaluation. The developed method contributes to more accurate evaluations of wellbore integrity for the oil and gas industry, leading to improved safety and environmental outcomes.","PeriodicalId":502250,"journal":{"name":"APL Machine Learning","volume":"26 13","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-04-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140739024","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Memristor crossbar arrays are expected to achieve highly energy-efficient neuromorphic computing via implementing parallel vector–matrix multiplication (VMM) in situ. The similarities between memristors and neural synapses offer opportunities for realizing hardware-based brain-inspired computing, such as spike neural networks. However, the nonlinear I–V characteristics of the memristors limit the implementation of parallel VMM on passive memristor crossbar arrays. In our work, we propose to utilize differential conductance as a synaptic weight to implement linear VMM operations on a passive memristor array in parallel. We fabricated a TiO2/HfO2 memristor crossbar array, in which differential-conductance-based synaptic weight exhibits plasticity, nonvolatility, multi-states, and tunable ON/OFF ratio. The noise-dependent accuracy performance of VMM operations based on the proposed approach was evaluated, offering an optimization guideline. Furthermore, we demonstrated a spike neural network circuit capable of processing small spiking signals through the differential-conductance-based synapses. The experimental results showcase effective space-coded and time-coded spike pattern recognition. Importantly, our work opens up new possibilities for the development of passive memristor arrays, leading to increased energy and area efficiency in brain-inspired chips.
{"title":"Harnessing nonlinear conductive characteristic of TiO2/HfO2 memristor crossbar for implementing parallel vector–matrix multiplication","authors":"Wei Wei, Cong Wang, Chen Pan, Xing-Jian Yangdong, Zaizheng Yang, Yuekun Yang, Bin Cheng, Shi-Jun Liang, Feng Miao","doi":"10.1063/5.0195190","DOIUrl":"https://doi.org/10.1063/5.0195190","url":null,"abstract":"Memristor crossbar arrays are expected to achieve highly energy-efficient neuromorphic computing via implementing parallel vector–matrix multiplication (VMM) in situ. The similarities between memristors and neural synapses offer opportunities for realizing hardware-based brain-inspired computing, such as spike neural networks. However, the nonlinear I–V characteristics of the memristors limit the implementation of parallel VMM on passive memristor crossbar arrays. In our work, we propose to utilize differential conductance as a synaptic weight to implement linear VMM operations on a passive memristor array in parallel. We fabricated a TiO2/HfO2 memristor crossbar array, in which differential-conductance-based synaptic weight exhibits plasticity, nonvolatility, multi-states, and tunable ON/OFF ratio. The noise-dependent accuracy performance of VMM operations based on the proposed approach was evaluated, offering an optimization guideline. Furthermore, we demonstrated a spike neural network circuit capable of processing small spiking signals through the differential-conductance-based synapses. The experimental results showcase effective space-coded and time-coded spike pattern recognition. Importantly, our work opens up new possibilities for the development of passive memristor arrays, leading to increased energy and area efficiency in brain-inspired chips.","PeriodicalId":502250,"journal":{"name":"APL Machine Learning","volume":"33 3","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-04-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140744586","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Carbon dioxide reduction is a major step toward building a cleaner and safer environment. There is a surge of interest in exploring high-entropy alloys (HEAs) as active catalysts for CO2 reduction; however, so far, it is mainly limited to quinary HEAs. Inspired by the successful synthesis of octonary and denary HEAs, herein, the CO2 reduction reaction (CO2RR) performance of an HEA composed of Ag, Au, Cu, Pd, Pt, Co, Ga, Ni, and Zn is studied by developing a high-fidelity graph neural network (GNN) framework. Within this framework, the adsorption site geometry and physics are employed through the featurization of elements. Particularly, featurization is performed using various intrinsic properties, such as electronegativity and atomic radius, to enable not only the supervised learning of CO2RR performance descriptors, namely, CO and H adsorption energies, but also the learning of adsorption physics and generalization to unseen metals and alloys. The developed model evaluates the adsorption strength of ∼3.5 and ∼0.4 billion possible sites for CO and H, respectively. Despite the enormous space of the AgAuCuPdPtCoGaNiZn alloy and the rather small size of the training data, the GNN framework demonstrated high accuracy and good robustness. This study paves the way for the rapid screening and intelligent synthesis of CO2RR-active and selective HEAs.
二氧化碳减排是建设更清洁、更安全环境的重要一步。人们对探索高熵合金（HEAs）作为二氧化碳还原活性催化剂的兴趣日益高涨；然而，迄今为止，这种研究主要局限于二元 HEAs。受成功合成八元和二元 HEA 的启发，本文通过开发高保真图神经网络 (GNN) 框架，研究了由 Ag、Au、Cu、Pd、Pt、Co、Ga、Ni 和 Zn 组成的 HEA 的二氧化碳还原反应 (CO2RR) 性能。在此框架内，通过元素的特征化，采用了吸附位点的几何形状和物理特性。特别是利用电负性和原子半径等各种固有属性进行特征化，不仅实现了 CO2RR 性能描述符（即 CO 和 H 吸附能）的监督学习，还实现了吸附物理学的学习以及对未知金属和合金的泛化。所开发的模型分别评估了 35 亿个和 4 亿个可能的 CO 和 H 吸附位点的吸附强度。尽管 AgAuCuPdPtCoGaNiZn 合金的空间巨大，而且训练数据的规模相当小，但 GNN 框架仍表现出很高的准确性和良好的鲁棒性。这项研究为快速筛选和智能合成具有 CO2RR 活性和选择性的 HEA 铺平了道路。
{"title":"Study of the adsorption sites of high entropy alloys for CO2 reduction using graph convolutional network","authors":"H. Oliaei, N. Aluru","doi":"10.1063/5.0198043","DOIUrl":"https://doi.org/10.1063/5.0198043","url":null,"abstract":"Carbon dioxide reduction is a major step toward building a cleaner and safer environment. There is a surge of interest in exploring high-entropy alloys (HEAs) as active catalysts for CO2 reduction; however, so far, it is mainly limited to quinary HEAs. Inspired by the successful synthesis of octonary and denary HEAs, herein, the CO2 reduction reaction (CO2RR) performance of an HEA composed of Ag, Au, Cu, Pd, Pt, Co, Ga, Ni, and Zn is studied by developing a high-fidelity graph neural network (GNN) framework. Within this framework, the adsorption site geometry and physics are employed through the featurization of elements. Particularly, featurization is performed using various intrinsic properties, such as electronegativity and atomic radius, to enable not only the supervised learning of CO2RR performance descriptors, namely, CO and H adsorption energies, but also the learning of adsorption physics and generalization to unseen metals and alloys. The developed model evaluates the adsorption strength of ∼3.5 and ∼0.4 billion possible sites for CO and H, respectively. Despite the enormous space of the AgAuCuPdPtCoGaNiZn alloy and the rather small size of the training data, the GNN framework demonstrated high accuracy and good robustness. This study paves the way for the rapid screening and intelligent synthesis of CO2RR-active and selective HEAs.","PeriodicalId":502250,"journal":{"name":"APL Machine Learning","volume":"749 ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-04-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140749153","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Perovskite-based optoelectronic devices have emerged as a promising energy source due to their potential for scalable production. This study introduces “perovskene,” a novel class of 2D materials derived from the ABC3-like perovskites, synthesized via a data-driven, high-throughput computational strategy. We harness machine learning and multitarget deep neural networks to systematically investigate the structure–property relations, paving the way for targeted material design and optimization in fields such as renewable energy, electronics, and catalysis. The characterization of over 1500 synthesized structures shows that more than 500 structures are stable, revealing properties such as ultra-low work function and large magnetic moment, underscoring the potential for advanced technological applications.
{"title":"Computational synthesis of a new generation of 2D-based perovskite quantum materials","authors":"C. Ekuma","doi":"10.1063/5.0189497","DOIUrl":"https://doi.org/10.1063/5.0189497","url":null,"abstract":"Perovskite-based optoelectronic devices have emerged as a promising energy source due to their potential for scalable production. This study introduces “perovskene,” a novel class of 2D materials derived from the ABC3-like perovskites, synthesized via a data-driven, high-throughput computational strategy. We harness machine learning and multitarget deep neural networks to systematically investigate the structure–property relations, paving the way for targeted material design and optimization in fields such as renewable energy, electronics, and catalysis. The characterization of over 1500 synthesized structures shows that more than 500 structures are stable, revealing properties such as ultra-low work function and large magnetic moment, underscoring the potential for advanced technological applications.","PeriodicalId":502250,"journal":{"name":"APL Machine Learning","volume":"24 10","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-04-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140753357","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Vehicle re-identification (V-ReID) is a critical task that aims to match the same vehicle across images from different camera viewpoints. The previous studies have leveraged attribute clues, such as color, model, and license plate, to enhance the V-ReID performance. However, these methods often lack effective interaction between the global–local features and the final V-ReID objective. Moreover, they do not address the challenging issues in real-world scenarios, such as high viewpoint variations, extreme illumination conditions, and car appearance changes (e.g., due to damage or wrong driving). We propose a novel framework to tackle these problems and advance the research in V-ReID, which can handle various types of car appearance changes and achieve robust V-ReID under varying lighting conditions. Our main contributions are as follows: (i) we propose a new Re-ID architecture named global–local self-attention network, which integrates local information into the feature learning process and enhances the feature representation for V-ReID and (ii) we introduce a novel damaged vehicle Re-ID dataset called VERI-D, which is the first publicly available dataset that focuses on this challenging yet practical scenario. The dataset contains both natural and synthetic images of damaged vehicles captured from multiple camera viewpoints and under different lighting conditions. (iii) We conduct extensive experiments on the VERI-D dataset and demonstrate the effectiveness of our approach in addressing the challenges associated with damaged vehicle re-identification. We also compare our method to several state-of-the-art V-ReID methods and show its superiority.
{"title":"VERI-D: A new dataset and method for multi-camera vehicle re-identification of damaged cars under varying lighting conditions","authors":"Shao Liu, S. Agaian","doi":"10.1063/5.0183408","DOIUrl":"https://doi.org/10.1063/5.0183408","url":null,"abstract":"Vehicle re-identification (V-ReID) is a critical task that aims to match the same vehicle across images from different camera viewpoints. The previous studies have leveraged attribute clues, such as color, model, and license plate, to enhance the V-ReID performance. However, these methods often lack effective interaction between the global–local features and the final V-ReID objective. Moreover, they do not address the challenging issues in real-world scenarios, such as high viewpoint variations, extreme illumination conditions, and car appearance changes (e.g., due to damage or wrong driving). We propose a novel framework to tackle these problems and advance the research in V-ReID, which can handle various types of car appearance changes and achieve robust V-ReID under varying lighting conditions. Our main contributions are as follows: (i) we propose a new Re-ID architecture named global–local self-attention network, which integrates local information into the feature learning process and enhances the feature representation for V-ReID and (ii) we introduce a novel damaged vehicle Re-ID dataset called VERI-D, which is the first publicly available dataset that focuses on this challenging yet practical scenario. The dataset contains both natural and synthetic images of damaged vehicles captured from multiple camera viewpoints and under different lighting conditions. (iii) We conduct extensive experiments on the VERI-D dataset and demonstrate the effectiveness of our approach in addressing the challenges associated with damaged vehicle re-identification. We also compare our method to several state-of-the-art V-ReID methods and show its superiority.","PeriodicalId":502250,"journal":{"name":"APL Machine Learning","volume":"19 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140271683","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
K. Burns, Kayvon Tadj, Tarun Allaparti, Liliana Arias, Nan Li, A. Aitkaliyeva, Amit Misra, M. Scott, Khalid Hattar
Modeling time-series data with convolutional neural networks (CNNs) requires building a model to learn in batches as opposed to training sequentially. Coupling CNNs with in situ or operando techniques opens the possibility of accurately segmenting dynamic reactions and mass transport phenomena to understand how materials behave under the conditions in which they are used. In this article, in situ ion irradiation transmission electron microscopy (TEM) images are used as inputs into the CNN to assess the defect generation rate, defect cluster density, and saturation of defects. We then use the output segmentation maps to correlate with conventional TEM micrographs to assess the model’s ability to detail nanoscale interactions. Next, we discuss the implications of preprocessing and hyperparameters on model variability, accuracy when expanded to other datasets, and the role of regularization when controlling model variance. Ultimately, we eliminate human bias when extrapolating physical metrics, speed up analysis time, decouple reactions that happen at 100 ms intervals, and deploy models that are both accurate and transferable to similar experiments.
利用卷积神经网络（CNN）对时间序列数据建模，需要建立一个分批学习的模型，而不是按顺序进行训练。将卷积神经网络与原位或操作技术相结合，可以准确地分割动态反应和质量传输现象，从而了解材料在使用条件下的行为。在本文中，原位离子照射透射电子显微镜（TEM）图像被用作 CNN 的输入，以评估缺陷生成率、缺陷群密度和缺陷饱和度。然后，我们使用输出分割图与传统 TEM 显微照片进行关联，以评估该模型详细描述纳米级相互作用的能力。接下来，我们讨论了预处理和超参数对模型变异性的影响、扩展到其他数据集时的准确性以及正则化在控制模型变异性时的作用。最终，我们消除了推断物理指标时的人为偏差，加快了分析时间，解耦了以 100 毫秒间隔发生的反应，并部署了既准确又可移植到类似实验的模型。
{"title":"Deep learning-enabled probing of irradiation-induced defects in time-series micrographs","authors":"K. Burns, Kayvon Tadj, Tarun Allaparti, Liliana Arias, Nan Li, A. Aitkaliyeva, Amit Misra, M. Scott, Khalid Hattar","doi":"10.1063/5.0186046","DOIUrl":"https://doi.org/10.1063/5.0186046","url":null,"abstract":"Modeling time-series data with convolutional neural networks (CNNs) requires building a model to learn in batches as opposed to training sequentially. Coupling CNNs with in situ or operando techniques opens the possibility of accurately segmenting dynamic reactions and mass transport phenomena to understand how materials behave under the conditions in which they are used. In this article, in situ ion irradiation transmission electron microscopy (TEM) images are used as inputs into the CNN to assess the defect generation rate, defect cluster density, and saturation of defects. We then use the output segmentation maps to correlate with conventional TEM micrographs to assess the model’s ability to detail nanoscale interactions. Next, we discuss the implications of preprocessing and hyperparameters on model variability, accuracy when expanded to other datasets, and the role of regularization when controlling model variance. Ultimately, we eliminate human bias when extrapolating physical metrics, speed up analysis time, decouple reactions that happen at 100 ms intervals, and deploy models that are both accurate and transferable to similar experiments.","PeriodicalId":502250,"journal":{"name":"APL Machine Learning","volume":"543 ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140280958","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Valtteri Torsti, T. Mäkinen, Silvia Bonfanti, J. Koivisto, Mikko J. Alava
The search for better compositions in high entropy alloys is a formidable challenge in materials science. Here, we demonstrate a systematic Bayesian optimization method to enhance the mechanical properties of the paradigmatic five-element Cantor alloy in silico. This method utilizes an automated loop with an online database, a Bayesian optimization algorithm, thermodynamic modeling, and molecular dynamics simulations. Starting from the equiatomic Cantor composition, our approach optimizes the relative fractions of its constituent elements, searching for better compositions while maintaining the thermodynamic phase stability. With 24 steps, we find Fe21Cr20Mn5Co20Ni34 with a yield stress improvement of 58%, and with 72 steps, we find Fe6Cr22Mn5Co32Ni35 where the yield stress has improved by 74%. These optimized compositions correspond to Ni-rich medium entropy alloys with enhanced mechanical properties and superior face-centered-cubic phase stability compared to the traditional equiatomic Cantor alloy. The automatic approach devised here paves the way for designing high entropy alloys with tailored properties, opening avenues for numerous potential applications.
{"title":"Improving the mechanical properties of Cantor-like alloys with Bayesian optimization","authors":"Valtteri Torsti, T. Mäkinen, Silvia Bonfanti, J. Koivisto, Mikko J. Alava","doi":"10.1063/5.0179844","DOIUrl":"https://doi.org/10.1063/5.0179844","url":null,"abstract":"The search for better compositions in high entropy alloys is a formidable challenge in materials science. Here, we demonstrate a systematic Bayesian optimization method to enhance the mechanical properties of the paradigmatic five-element Cantor alloy in silico. This method utilizes an automated loop with an online database, a Bayesian optimization algorithm, thermodynamic modeling, and molecular dynamics simulations. Starting from the equiatomic Cantor composition, our approach optimizes the relative fractions of its constituent elements, searching for better compositions while maintaining the thermodynamic phase stability. With 24 steps, we find Fe21Cr20Mn5Co20Ni34 with a yield stress improvement of 58%, and with 72 steps, we find Fe6Cr22Mn5Co32Ni35 where the yield stress has improved by 74%. These optimized compositions correspond to Ni-rich medium entropy alloys with enhanced mechanical properties and superior face-centered-cubic phase stability compared to the traditional equiatomic Cantor alloy. The automatic approach devised here paves the way for designing high entropy alloys with tailored properties, opening avenues for numerous potential applications.","PeriodicalId":502250,"journal":{"name":"APL Machine Learning","volume":"269 ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140275165","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Inverse design is an application of machine learning to device design, giving the computer maximal latitude in generating novel structures, learning from their performance, and optimizing them to suit the designer’s needs. Gradient-based optimizers, augmented by the adjoint method to efficiently compute the gradient, are particularly attractive for this approach and have proven highly successful with finite-element and finite-difference physics simulators. Here, we extend adjoint optimization to the transfer matrix method, an accurate and efficient simulator for a wide variety of quasi-1D physical phenomena. We leverage this versatility to develop a physics-agnostic inverse design framework and apply it to three distinct problems, each presenting a substantial challenge for conventional design methods: optics, designing a multivariate optical element for compressive sensing; acoustics, designing a high-performance anti-sonar submarine coating; and quantum mechanics, designing a tunable double-bandpass electron energy filter.
{"title":"Physics-agnostic inverse design using transfer matrices","authors":"Nathaniel Morrison, Shuaiwei Pan, Eric Y. Ma","doi":"10.1063/5.0179457","DOIUrl":"https://doi.org/10.1063/5.0179457","url":null,"abstract":"Inverse design is an application of machine learning to device design, giving the computer maximal latitude in generating novel structures, learning from their performance, and optimizing them to suit the designer’s needs. Gradient-based optimizers, augmented by the adjoint method to efficiently compute the gradient, are particularly attractive for this approach and have proven highly successful with finite-element and finite-difference physics simulators. Here, we extend adjoint optimization to the transfer matrix method, an accurate and efficient simulator for a wide variety of quasi-1D physical phenomena. We leverage this versatility to develop a physics-agnostic inverse design framework and apply it to three distinct problems, each presenting a substantial challenge for conventional design methods: optics, designing a multivariate optical element for compressive sensing; acoustics, designing a high-performance anti-sonar submarine coating; and quantum mechanics, designing a tunable double-bandpass electron energy filter.","PeriodicalId":502250,"journal":{"name":"APL Machine Learning","volume":"43 5","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-02-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140421641","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
M. Stern, Sam Dillavou, Dinesh Jayaraman, D. Durian, Andrea J. Liu
As the size and ubiquity of artificial intelligence and computational machine learning models grow, the energy required to train and use them is rapidly becoming economically and environmentally unsustainable. Recent laboratory prototypes of self-learning electronic circuits, such as “physical learning machines,” open the door to analog hardware that directly employs physics to learn desired functions from examples at a low energy cost. In this work, we show that this hardware platform allows for an even further reduction in energy consumption by using good initial conditions and a new learning algorithm. Using analytical calculations, simulations, and experiments, we show that a trade-off emerges when learning dynamics attempt to minimize both the error and the power consumption of the solution—greater power reductions can be achieved at the cost of decreasing solution accuracy. Finally, we demonstrate a practical procedure to weigh the relative importance of error and power minimization, improving the power efficiency given a specific tolerance to error.
{"title":"Training self-learning circuits for power-efficient solutions","authors":"M. Stern, Sam Dillavou, Dinesh Jayaraman, D. Durian, Andrea J. Liu","doi":"10.1063/5.0181382","DOIUrl":"https://doi.org/10.1063/5.0181382","url":null,"abstract":"As the size and ubiquity of artificial intelligence and computational machine learning models grow, the energy required to train and use them is rapidly becoming economically and environmentally unsustainable. Recent laboratory prototypes of self-learning electronic circuits, such as “physical learning machines,” open the door to analog hardware that directly employs physics to learn desired functions from examples at a low energy cost. In this work, we show that this hardware platform allows for an even further reduction in energy consumption by using good initial conditions and a new learning algorithm. Using analytical calculations, simulations, and experiments, we show that a trade-off emerges when learning dynamics attempt to minimize both the error and the power consumption of the solution—greater power reductions can be achieved at the cost of decreasing solution accuracy. Finally, we demonstrate a practical procedure to weigh the relative importance of error and power minimization, improving the power efficiency given a specific tolerance to error.","PeriodicalId":502250,"journal":{"name":"APL Machine Learning","volume":"55 7","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-02-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140427750","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：