Big data and machine learning for materials science

Abstract

Herein, we review aspects of leading-edge research and innovation in materials science that exploit big data and machine learning (ML), two computer science concepts that combine to yield computational intelligence. ML can accelerate the solution of intricate chemical problems and even solve problems that otherwise would not be tractable. However, the potential benefits of ML come at the cost of big data production; that is, the algorithms demand large volumes of data of various natures and from different sources, from material properties to sensor data. In the survey, we propose a roadmap for future developments with emphasis on computer-aided discovery of new materials and analysis of chemical sensing compounds, both prominent research fields for ML in the context of materials science. In addition to providing an overview of recent advances, we elaborate upon the conceptual and practical limitations of big data and ML applied to materials science, outlining processes, discussing pitfalls, and reviewing cases of success and failure.

Sensing and Biosensing in the World of Autonomous Machines and Intelligent Systems

“…The various issues involved in dealing with big data and machine learning for applications such as the autonomous station have been discussed in reviews and opinion papers (Oliveira et al, 2014;Rodrigues et al, 2016;Paulovich et al, 2018;Rodrigues et al, 2021). We observe a synergistic movement driven by the combination of data generation at unprecedented levels of detail, variety and velocity with massive computing capability.…”

“…The rapid progress in autonomous systems with artificial intelligence (AI) has brought an expectation that machines and software systems will soon be able to perform intellectual tasks as efficiently as humans (perhaps even better), to the extent that in a near future, for the first time in history such systems may be generating knowledge, with no human intervention (Rodrigues et al, 2021). This tremendous achievement will only be realized if these manmade systems can acquire, process and make sense of a lot of combined data from the environment, in addition to mastering natural languages.…”

“…This tremendous achievement will only be realized if these manmade systems can acquire, process and make sense of a lot of combined data from the environment, in addition to mastering natural languages. The latter requirement appears particularly challenging since machine learning (ML) and other currently successful AI approaches are not yet sufficient to interpret text (Rodrigues et al, 2021). Another stringent requirement is in the capability of continuously acquiring information with sensors and biosensors, in many cases having to emulate human capabilities.…”

High throughput inverse design and Bayesian optimization of functionalities: spin splitting in two-dimensional compounds

“…1 Although this process has been successfully implemented for functionalities such as ferroelectricity [2][3][4] and two-dimensional materials [5][6][7] , this direct approach is usually tedious and expensive. The second step consists in the extrapolations of numerical correlations found with approaches such as machine learning [8][9][10][11][12] or cluster expansion methods 13 . However, numerical relations are not necessarily transferable (i.e., limited to the set of compounds used to train the modelstraining set), preventing the rational design of material candidates with the optimized property out of the training set.…”

Tailoring Vibrational Signature and Functionality of 2D-Ordered Linear-Chain Carbon-Based Nanocarriers for Predictive Performance Enhancement of High-End Energetic Materials

“…Nowadays, research on nanomaterials science is rapidly entering the phase of a data-driven age. For the predictive growth of the 2D-ordered linear-chain carbon-based functionalizing nanocarriers with a unique set of programmable microarchitectures and physicochemical properties, by using the extensive experimental testing, we propose to apply a new paradigm in materials science—a science based on data and deep materials informatics [ 53 , 54 , 55 ]. In this research area, the experimental data is a new resource, and knowledge is extracted from the datasets of materials.…”