Machine learning-based inverse design methods considering data characteristics and design space size in materials design and manufacturing: a review

Lee, Junhyeong; Park, Dong-Geun; Lee, Mingyu; Lee, Hugon; Park, Kundo; Lee, Ikjin; Ryu, Seunghwa

doi:10.1039/d3mh00039g

Cited by 22 publications

(2 citation statements)

References 139 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Machine learning (ML) has recently become widely recognized and influential across various scientific disciplines, including the field of materials science, [8][9][10][11][12]. These approaches focused on data have greatly expedited the process of exploring and developing innovative materials.…”

Section: The Fusion Of Materials Science Tetrahedron Paradigm and Dee...mentioning

confidence: 99%

Unlocking the Hidden Potential of the Data-Driven Nanocarbon Genome: Unleashing Novel Neural Pathways, Elucidating Growth-Property Relationships, and Enabling Inverse Design

Lukin,

Gülseren

2024

Preprint

View full text Add to dashboard Cite

The swift progress in low-dimensional nanocarbons, specifically in the realm of the authentic one-dimensional carbon chain featuring sp1 hybridization, has unveiled exciting prospects for integrating them into cutting-edge technologies across diverse industries. To fully harness their potential, precise control over the growth process is necessary to obtain desired nanostructure and functionality. However, optimizing properties through traditional approaches has been challenging due to complex interactions. To address this, we implement a focused data-driven strategy for nanocarbon inverse design by leveraging a state-of-the-art data-driven nanocarbon genome approach (NCGA). By uncovering relationships between growth parameters and resultant traits, this serves as an expedient catalyst in engineering nanostructures with tailored attributes. We introduce an extensive array of technological approaches aimed at precisely controlling the growth process. These methods encompass stimulating and precisely adjusting synergistic effects, coordinating atomic vibrations at a collective level through the implementation of multilayer nano-interfaces, harnessing the potential of active screen plasma surface engineering, utilizing nano-patterning and allotropic phase transformations, incorporating heteroatom doping, and effectively directing self-assembly. These aim to unlock nanomaterials' latent potential and reveal novel neural pathways within the data-driven NCGA, enhancing its predictive capabilities. Specifically, triggering self-organization during growth can potentially unlock previously unexplored neural pathways. The data-driven NCGA offers a paradigm shift, accelerating discovery and design of nanocarbons with optimized, application-specific properties.

show abstract

Section: The Fusion Of Materials Science Tetrahedron Paradigm and Dee...mentioning

confidence: 99%

Unlocking the Hidden Potential of the Data-Driven Nanocarbon Genome: Unleashing Novel Neural Pathways, Elucidating Growth-Property Relationships, and Enabling Inverse Design

Lukin,

Gülseren

2024

Preprint

View full text Add to dashboard Cite

show abstract

“…Further evidence of the importance of inverse problems lies in the growing interest in combining them with design strategies [7][8][9][10][11]. For example, imagine that we wish to build a hypothetical physical system that responds to external stimuli in a certain way.…”

Section: Introductionmentioning

confidence: 99%

Inverse design from the catenary problem

de Lima,

de Araújo,

Ferreira

2024

Eur. J. Phys.

View full text Add to dashboard Cite

Inverse problems in science normally involve the challenge of obtaining from a set of observations the causal factors that generated them in the first place. However, physics students are seldom exposed to such problems as part of their training. Here we revisit the mechanics problem of finding the shape of a hanging cable, but this time in reverse, {\it i.e.}, by asking what mass density a cable must have to follow a specific shape. This concept is then generalised into the possibility of identifying a cable whose hanging shape follows any form we wish to design. This inverted design strategy is experimentally verified with an inexpensive setup that is suitable for classroom activities.

show abstract

Electrode Placement Optimization for Electrical Impedance Tomography Using Active Learning

Lee,

Park,

Park

et al. 2024

Adv Eng Mater

Self Cite

View full text Add to dashboard Cite

Electrical Impedance Tomography (EIT) offers a versatile imaging modality with a multitude of applications, although it encounters accuracy limitations. In this study, we present a novel systematic framework that integrates a neural network (NN), active learning, and transfer learning to optimize electrode placement, improving image reconstruction performance based on user‐defined metrics. Given the many‐to‐one mapping between electrode configuration and the performance metric, our approach utilizes a NN that predicts performance metrics from electrode placement input. To maintain NN’s prediction accuracy for unseen electrode configurations, we maximize performance metrics while iteratively updating the neural network via active learning during the optimization process. We employ transfer learning to expedite optimization of electrode placements for time‐consuming iterative reconstruction techniques by fine‐tuning a NN initially trained on one‐step reconstruction data. We validate our method using two representative reconstruction methods: one‐step reconstruction with Newton’s one‐step error reconstructor (NOSER) prior and the iterative Total Variation (TV) method. This research underscores the potential of our proposed framework in addressing EIT’s inherent limitations and augmenting its performance across diverse applications and reconstruction methods. The framework could potentially contribute to the advancement of non‐invasive medical imaging, structural health monitoring, strain sensing, robotics, and other fields that depend on EIT.This article is protected by copyright. All rights reserved.

show abstract

Machine learning-based inverse design methods considering data characteristics and design space size in materials design and manufacturing: a review

Abstract: In the last few decades, the influence of machine learning has permeated many areas of science and technology, including the field of material science. This toolkit of statistical methods accelerated...

Cited by 22 publications

References 139 publications

Unlocking the Hidden Potential of the Data-Driven Nanocarbon Genome: Unleashing Novel Neural Pathways, Elucidating Growth-Property Relationships, and Enabling Inverse Design

Unlocking the Hidden Potential of the Data-Driven Nanocarbon Genome: Unleashing Novel Neural Pathways, Elucidating Growth-Property Relationships, and Enabling Inverse Design

Inverse design from the catenary problem

Electrode Placement Optimization for Electrical Impedance Tomography Using Active Learning

Contact Info

Product

Resources

About