Explainable machine learning in materials science (2022-09-22T00:00:00.000000Z)

TL;DR

This article attempts to provide an entry point to explainable artificial intelligence (XAI), an emerging research field that addresses the explainability of complicated machine learning models like deep neural networks (DNNs).

Abstract

Machine learning models are increasingly used in materials studies because of their exceptional accuracy. However, the most accurate machine learning models are usually difficult to explain. Remedies to this problem lie in explainable artificial intelligence (XAI), an emerging research field that addresses the explainability of complicated machine learning models like deep neural networks (DNNs). This article attempts to provide an entry point to XAI for materials scientists. Concepts are defined to clarify what explain means in the context of materials science. Example works are reviewed to show how XAI helps materials science research. Challenges and opportunities are also discussed.

Authors

Xiaoting Zhong

1 papers

Brian Gallagher

2 papers

Shusen Liu

1 papers

TL;DR

Abstract

Authors

References160 items

Efficient and interpretable graph network representation for angle-dependent properties applied to optical spectroscopy

Attribution-Driven Explanation of the Deep Neural Network Model via Conditional Microstructure Image Synthesis

Machine-learning-based intelligent framework for discovering refractory high-entropy alloys with improved high-temperature yield strength

Notions of explainability and evaluation approaches for explainable artificial intelligence

Learning from learning machines: a new generation of AI technology to meet the needs of science

Interpretable and Explainable Machine Learning for Materials Science and Chemistry

A machine learning-based alloy design system to facilitate the rational design of high entropy alloys with enhanced hardness

A study of real-world micrograph data quality and machine learning model robustness

SHAP and LIME: An Evaluation of Discriminative Power in Credit Risk

Non-intrusive reduced order modeling of natural convection in porous media using convolutional autoencoders: comparison with linear subspace techniques

Highly accurate protein structure prediction with AlphaFold

Reliable Graph Neural Network Explanations Through Adversarial Training

Machine learning in materials science: From explainable predictions to autonomous design

Physics-informed machine learning

Leveraging Uncertainty from Deep Learning for Trustworthy Material Discovery Workflows

Informing Mechanical Model Development Using Lower-Dimensional Descriptions of Lattice Distortion

Interpretable machine-learning strategy for soft-magnetic property and thermal stability in Fe-based metallic glasses

Efficient nonlinear manifold reduced order model

Application of a long short-term memory for deconvoluting conductance contributions at charged ferroelectric domain walls

An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale

Quantitative interpretation explains machine learning models for chemical reaction prediction and uncovers bias

Nonlinear manifold-based reduced order model

A fast and accurate physics-informed neural network reduced order model with shallow masked autoencoder

Better, Faster, and Less Biased Machine Learning: Electromechanical Switching in Ferroelectric Thin Films

Mapping the space of chemical reactions using attention-based neural networks

Dramatically Enhanced Combination of Ultimate Tensile Strength and Electric Conductivity of Alloys via Machine Learning Screening

Overview: Computer Vision and Machine Learning for Microstructural Characterization and Analysis

Benchmark AFLOW Data Sets for Machine Learning

The Best Way to Select Features? Comparing MDA, LIME, and SHAP

Machine-learning informed prediction of high-entropy solid solution formation: Beyond the Hume-Rothery rules

State-of-the-art augmented NLP transformer models for direct and single-step retrosynthesis

Learning acoustic emission signatures from a nanoindentation-based lithography process: Towards rapid microstructure characterization

High-throughput calculations of magnetic topological materials

Compositionally restricted attention-based network for materials property predictions

Multi‐Layer Feature Selection Incorporating Weighted Score‐Based Expert Knowledge toward Modeling Materials with Targeted Properties

Identification of advanced spin-driven thermoelectric materials via interpretable machine learning

Deep Learning for Optoelectronic Properties of Organic Semiconductors

Explainable Artificial Intelligence (XAI): Concepts, Taxonomies, Opportunities and Challenges toward Responsible AI

One Explanation Does Not Fit All: A Toolkit and Taxonomy of AI Explainability Techniques

Recent advances and applications of machine learning in solid-state materials science

Discovery of new materials using combinatorial synthesis and high-throughput characterization of thin-film materials libraries combined with computational methods

Predicting compressive strength of consolidated molecular solids using computer vision and deep learning

MatCALO: Knowledge-enabled machine learning in materials science

Explainability Methods for Graph Convolutional Neural Networks

Explainable Machine Learning for Scientific Insights and Discoveries

Four-Dimensional Scanning Transmission Electron Microscopy (4D-STEM): From Scanning Nanodiffraction to Ptychography and Beyond

Adversarial Examples Are Not Bugs, They Are Features

Attention Augmented Convolutional Networks

Informed Machine Learning – A Taxonomy and Survey of Integrating Prior Knowledge into Learning Systems

Establishing structure-property localization linkages for elastic deformation of three-dimensional high contrast composites using deep learning approaches

Machine-learning guided discovery of a new thermoelectric material

A physics-aware, probabilistic machine learning framework for coarse-graining high-dimensional systems in the Small Data regime

Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations

Reliable and explainable machine-learning methods for accelerated material discovery

ElemNet: Deep Learning the Chemistry of Materials From Only Elemental Composition

Fast and interpretable classification of small X-ray diffraction datasets using data augmentation and deep neural networks

Molecular Transformer: A Model for Uncertainty-Calibrated Chemical Reaction Prediction

Sanity Checks for Saliency Maps

Peeking Inside the Black-Box: A Survey on Explainable Artificial Intelligence (XAI)

SchNetPack: A Deep Learning Toolbox For Atomistic Systems.

Microstructure Cluster Analysis with Transfer Learning and Unsupervised Learning

A Survey on Deep Transfer Learning

Using machine learning and a data-driven approach to identify the small fatigue crack driving force in polycrystalline materials

Physics-aware Gaussian processes in remote sensing

This looks like that: deep learning for interpretable image recognition

Quantum-chemical insights from interpretable atomistic neural networks

On the Robustness of Interpretability Methods

Towards Robust Interpretability with Self-Explaining Neural Networks

Advanced microstructure classification by data mining methods

Explaining Explanations: An Overview of Interpretability of Machine Learning

A Comparative Study of Feature Selection Methods for Stress Hotspot Classification in Materials

Material structure-property linkages using three-dimensional convolutional neural networks

Local Explanation Methods for Deep Neural Networks Lack Sensitivity to Parameter Values

A Survey of Methods for Explaining Black Box Models

SchNet - A deep learning architecture for molecules and materials.

Microstructure recognition using convolutional neural networks for prediction of ionic conductivity in ceramics