Search

Machine-learning (ML)-based interatomic potentials are increasingly popular in material modeling, enabling highly accurate simulations with thousands and millions of atoms. However, the performance of machine-learned potentials depends strongly on the choice of hyperparameters—that is, of those parameters that are set before the model encounters data. This problem is particularly acute where hyperparameters have no intuitive physical interpretation and where the corresponding optimization space is large. Here, we describe an openly available Python package that facilitates hyperparameter optimization across different ML potential fitting frameworks. We discuss methodological aspects relating to the optimization itself and to the selection of validation data, and we show example applications. We expect this package to become part of a wider computational framework to speed up the mainstream adaptation of ML potentials in the physical sciences. [ABSTRACT FROM AUTHOR]

The text discusses various developments in the fields of biotechnology, pharmaceuticals, chemicals, and other industries. It highlights the launch of the International Centre for Enzyme Design in the UK, partnerships between companies for cancer treatment research, and merger agreements in the ophthalmic pharma sector. It also covers collaborations for more sustainable chemical production, the establishment of a new injectables manufacturing facility, and the sale of a business unit. The text provides information on investments, expansions, collaborations, new facilities, and acquisitions by various companies. Additionally, it mentions the approval of a German measure to support a semiconductor manufacturing company and the relocation of a company. [Extracted from the article]

This document is a compilation of various scientific papers and preprints covering a wide range of topics, including protein folding, cell migration, machine learning, deep neural networks, neural scaling laws, representations and generalization in artificial and brain neural networks, and the neuron as a direct data-driven controller. The papers discuss different aspects of these subjects, providing a comprehensive overview of the current research in these fields. This document can be a valuable resource for library patrons conducting research on these topics, particularly in the fields of neuroscience and artificial intelligence. [Extracted from the article]

In recent years, metasurface, as a representative of micro- and nano-optics, have demonstrated a powerful ability to manipulate light, which can modulate a variety of physical parameters, such as wavelength, phase, and amplitude, to achieve various functions and substantially improve the performance of conventional optical components and systems. Artificial Intelligence (AI) is an emerging strong and effective computational tool that has been rapidly integrated into the study of physical sciences over the decades and has played an important role in the study of metasurface. This review starts with a brief introduction to the basics and then describes cases where AI and metasurface research have converged: from AI-assisted design of metasurface elements up to advanced optical systems based on metasurface. We demonstrate the advanced computational power of AI, as well as its ability to extract and analyze a wide range of optical information, and analyze the limitations of the available research resources. Finally conclude by presenting the challenges posed by the convergence of disciplines. [ABSTRACT FROM AUTHOR]

Introduction: This study presents the application of machine learning (ML) to evaluate marine fog visibility conditions and nowcasting of visibility based on the FATIMA (Fog and turbulence interactions in the marine atmosphere) campaign observations collected during July 2022 in the North Atlantic in the Grand Banks area and vicinity of Sable Island, northeast of Canada. Methods: The measurements were collected using instrumentation mounted on the Research Vessel Atlantic Condor. The collected meteorological parameters were: visibility (Vis), precipitation rate, air temperature, relative humidity with respect to water, pressure, wind speed, and direction. Using all variables, the droplet number concentration was used to qualitatively indicate and assess characteristics of the fog using the t-distributed stochastic neighbor embedding projection method (t-SNE), which clustered the data into groups. Following t-SNE analysis, a correlation heatmap was used to select relevant meteorological variables for visibility nowcasting, which were wind speed, relative humidity, and dew point depression. Prior to nowcasting, the input variables were preprocessed to generate additional time-lagged variables using a 120-minute lookback window in order to take advantage of the intrinsic time-varying features of the time series data. Nowcasting of Vis time series for lead times of 30 and 60 minutes was performed using the ML regression methods of support vector regression (SVR), least-squares gradient boosting (LSB), and deep learning at visibility thresholds of Vis < 1 km and < 10 km. Results: Vis nowcasting at the 60 min lead time was best with LSB and was significantly more skillful than persistence analysis. Specifically, using LSB the overall nowcasts at Vis 1 < km and Vis 10 < km were RMSE = 0.172 km and RMSE = 2.924 km, respectively. The nowcasting skill of SVR for dense fog (Vis ≤ 400 m) was significantly better than persistence at all Vis thresholds and lead times, even when it was less skillful than persistence at predicting high visibility. Discussion: Thus, ML techniques can significantly improve Vis prediction when either observations or modelbased accurate time-dependent variables are available. The results suggest that there is potential for future ML analysis that focuses on modeling the underlying factors of fog formation. [ABSTRACT FROM AUTHOR]

Manifold visualisation techniques are commonly used to visualise high-dimensional datasets in physical sciences. In this paper, we apply a recently introduced manifold visualisation method, slisemap, on datasets from physics and chemistry. slisemap combines manifold visualisation with explainable artificial intelligence. Explainable artificial intelligence investigates the decision processes of black box machine learning models and complex simulators. With slisemap, we find an embedding such that data items with similar local explanations are grouped together. Hence, slisemap gives us an overview of the different behaviours of a black box model, where the patterns in the embedding reflect a target property. In this paper, we show how slisemap can be used and evaluated on physical data and that it is helpful in finding meaningful information on classification and regression models trained on these datasets. [ABSTRACT FROM AUTHOR]

Introduction: Advances in machine learning (ML) methodologies, combined with multidisciplinary collaborations across biological and physical sciences, has the potential to propel drug discovery and development. Open Science fosters this collaboration by releasing datasets and methods into the public space; however, further education and widespread acceptance and adoption of Open Science approaches are necessary to tackle the plethora of known disease states. Motivation: In addition to providing much needed insights into potential therapeutic protein targets, we also aim to demonstrate that small patient datasets have the potential to provide insights that usually require many samples (>5,000). There are many such datasets available and novel advancements in ML can provide valuable insights from these patient datasets. Problem statement: Using a public dataset made available by patient advocacy group AnswerALS and a multidisciplinary Open Science approach with a systems biology augmented ML technology, we aim to validate previously reported drug targets in ALS and provide novel insights about ALS subpopulations and potential drug targets using a unique combination of ML methods and graph theory. Methodology: We use NetraAI to generate hypotheses about specific patient subpopulations, which were then refined and validated through a combination of ML techniques, systems biology methods, and expert input. Results: We extracted 8 target classes, each comprising of several genes that shed light into ALS pathophysiology and represent new avenues for treatment. These target classes are broadly categorized as inflammation, epigenetic, heat shock, neuromuscular junction, autophagy, apoptosis, axonal transport, and excitotoxicity. These findings are not mutually exclusive, and instead represent a systematic view of ALS pathophysiology. Based on these findings, we suggest that simultaneous targeting of ALS has the potential to mitigate ALS progression, with the plausibility of maintaining and sustaining an improved quality of life (QoL) for ALS patients. Even further, we identified subpopulations based on disease onset. Conclusion: In the spirit of Open Science, this work aims to bridge the knowledge gap in ALS pathophysiology to aid in diagnostic, prognostic, and therapeutic strategies and pave the way for the development of personalized treatments tailored to the individual's needs. [ABSTRACT FROM AUTHOR]

This document is an editorial for a special issue of the journal "Technologies" focused on data science and big data in various fields such as biology, physical science, and engineering. The editorial highlights the importance of analyzing large amounts of data generated by digital technologies and the need for data scientists to use artificial intelligence and machine learning to extract valuable knowledge. The special issue includes 12 papers covering topics such as machine learning techniques for customer churn prediction, agile program management in the U.S. Navy, deep learning for cybersecurity in Industry 5.0, self-directed learning during the COVID-19 era, decision tree-based neural networks for data classification, data-driven governance in technology companies, and more. The papers explore different approaches, models, and tools in the context of data science and big data. [Extracted from the article]

This article explores the use of Neural Network Potentials (NNPs) to model prebiotic chemistries accurately and efficiently. The authors present a scalable and generalizable approach for designing NNPs that can handle chemical reactivity in solvated systems. Through active learning and enhanced sampling techniques, they generate free-energy landscapes and calculate committors, finding a preference for the dissociative mechanism and identifying HPO4 2− as the more reactive species under prebiotic conditions. This research has the potential to enhance our understanding of prebiotic chemistry and its connection to the origins of life. The study acknowledges the need for further improvements, such as incorporating machine learning approaches and training neural network potentials using collective variables. The work was supported by the US Department of Energy. [Extracted from the article]

Physics is a field of science that has traditionally used the scientific method to answer questions about why natural phenomena occur and to make testable models that explain the phenomena. Discovering equations, laws, and principles that are invariant, robust, and causal has been fundamental in physical sciences throughout the centuries. Discoveries emerge from observing the world and, when possible, performing interventions on the system under study. With the advent of big data and data-driven methods, the fields of causal and equation discovery have developed and accelerated progress in computer science, physics, statistics, philosophy, and many applied fields. This paper reviews the concepts, methods, and relevant works on causal and equation discovery in the broad field of physics and outlines the most important challenges and promising future lines of research. We also provide a taxonomy for data-driven causal and equation discovery, point out connections, and showcase comprehensive case studies in Earth and climate sciences, fluid dynamics and mechanics, and the neurosciences. This review demonstrates that discovering fundamental laws and causal relations by observing natural phenomena is revolutionised with the efficient exploitation of observational data and simulations, modern machine learning algorithms and the combination with domain knowledge. Exciting times are ahead with many challenges and opportunities to improve our understanding of complex systems. [ABSTRACT FROM AUTHOR]

Large language models (LLMs), such as ChatGPT and PaLM, are able to perform sophisticated text comprehension and generation tasks with little or no training. Alongside their broader societal impacts, these capabilities carry great promise for the physical sciences, including applied mechanics. We present a summary of recent developments in these models, their application to mechanics and adjacent fields, and a perspective on their future use in applied mechanics, taking into account their limitations and the unique challenges of the field. [ABSTRACT FROM AUTHOR]

Noise and uncertainty are usually the enemy of machine learning, noise in training data leads to uncertainty and inaccuracy in the predictions. However, we develop a machine learning architecture that extracts crucial information out of the noise itself to improve the predictions. The phenomenology computes and then utilizes uncertainty in one target variable to predict a second target variable. We apply this formalism to PbZr0.7Sn0.3O3 crystal, using the uncertainty in dielectric constant to extrapolate heat capacity, correctly predicting a phase transition that otherwise cannot be extrapolated. For the second example – single-particle diffraction of droplets – we utilize the particle count together with its uncertainty to extrapolate the ground truth diffraction amplitude, delivering better predictions than when we utilize only the particle count. Our generic formalism enables the exploitation of uncertainty in machine learning, which has a broad range of applications in the physical sciences and beyond. [ABSTRACT FROM AUTHOR]

Jonathan Hopkins, CC BY-ND The experimental study lays the foundation for AI materials that can be applied in the construction of buildings, airplanes, and imaging technologies. For more information: https://engineering.lehigh.edu/news/article/cooler-approach-making-new-materials-can-stand-heat AI Capable of Predicting Properties of Multifaceted Metamaterials In a new study by the University of Amsterdam's Institute of Physics and research institute AMOLF, the researchers tested how well artificial intelligence (AI) can estimate the properties of so-called combinatorial mechanical metamaterials. Courtesy of A. Kuzmin/University of Latvia and A. Smekhova/HZB To investigate the local environment of individual components, the team used multi-edge x-ray absorption spectroscopy (EXAFS) at BESSY II and then the reverse Monte Carlo method to analyze the collected data. High-entropy alloys are a new class of alloys that are composed of four or more metallic elements in approximately equal amounts. [Extracted from the article]

Physics‐driven discovery in an autonomous experiment has emerged as a dream application of machine learning in physical sciences. Here, this work develops and experimentally implements a deep kernel learning (DKL) workflow combining the correlative prediction of the target functional response and its uncertainty from the structure, and physics‐based selection of acquisition function, which autonomously guides the navigation of the image space. Compared to classical Bayesian optimization (BO) methods, this approach allows to capture the complex spatial features present in the images of realistic materials, and dynamically learn structure–property relationships. In combination with the flexible scalarizer function that allows to ascribe the degree of physical interest to predicted spectra, this enables physical discovery in automated experiment. Here, this approach is illustrated for nanoplasmonic studies of nanoparticles and experimentally implemented in a truly autonomous fashion for bulk‐ and edge plasmon discovery in MnPS3, a lesser‐known beam‐sensitive layered 2D material. This approach is universal, can be directly used as‐is with any specimen, and is expected to be applicable to any probe‐based microscopic techniques including other STEM modalities, scanning probe microscopies, chemical, and optical imaging. [ABSTRACT FROM AUTHOR]

More Efficient Bayesian Optimization Through the Use of Common Random Numbers Bayesian optimization is a powerful tool for expensive stochastic black-box optimization problems, such as simulation-based optimization or hyperparameter tuning in machine learning systems. In "Bayesian Optimization Allowing for Common Random Numbers," Pearce, Poloczek, and Branke show how explicitly modeling the random seed in the Gaussian process surrogate model allows Bayesian optimization to exploit the structure in the noise and benefit from variance reduction provided by common random numbers. The proposed knowledge gradient with common random numbers acquisition function iteratively determines a combination of input and random seed to evaluate the objective. It automatically trades off reusing old seeds to benefit from the variance reduction through common random numbers and querying new seeds to avoid bias because of a small number of seeds. The proposed algorithm is analyzed theoretically and empirically shows superior performance compared with previous approaches on various test problems. Bayesian optimization is a powerful tool for expensive stochastic black-box optimization problems, such as simulation-based optimization or machine learning hyperparameter tuning. Many stochastic objective functions implicitly require a random number seed as input. By explicitly reusing a seed, a user can exploit common random numbers, comparing two or more inputs under the same randomly generated scenario, such as a common customer stream in a job shop problem or the same random partition of training data into training and validation sets for a machine learning algorithm. With the aim of finding an input with the best average performance over infinitely many seeds, we propose a novel Gaussian process model that jointly models both the output for each seed and the average over seeds. We then introduce the knowledge gradient for common random numbers that iteratively determines a combination of input and a random seed to evaluate the objective and automatically trades off reusing old seeds and querying new seeds, thus overcoming the need to evaluate inputs in batches or measure differences of pairs as suggested in previous methods. We investigate the knowledge gradient for common random numbers both theoretically and empirically, finding that it achieves significant performance improvements with only moderate added computational cost. Funding: The first author gratefully acknowledges funding through the UK Engineering and Physical Sciences Research Council [Grant EP/101358X/1]. Supplemental Material: The e-companion is available at https://doi.org/10.1287/opre.2021.2208. [ABSTRACT FROM AUTHOR]

Machine learning (ML) has emerged as a formidable force for identifying hidden but pertinent patterns within a given data set with the objective of subsequent generation of automated predictive behavior. In recent years, it is safe to conclude that ML and its close cousin, deep learning (DL), have ushered in unprecedented developments in all areas of physical sciences, especially chemistry. Not only classical variants of ML, even those trainable on near-term quantum hardwares have been developed with promising outcomes. Such algorithms have revolutionized materials design and performance of photovoltaics, electronic structure calculations of ground and excited states of correlated matter, computation of force-fields and potential energy surfaces informing chemical reaction dynamics, reactivity inspired rational strategies of drug designing and even classification of phases of matter with accurate identification of emergent criticality. In this review we shall explicate a subset of such topics and delineate the contributions made by both classical and quantum computing enhanced machine learning algorithms over the past few years. We shall not only present a brief overview of the well-known techniques but also highlight their learning strategies using statistical physical insight. The objective of the review is not only to foster exposition of the aforesaid techniques but also to empower and promote cross-pollination among future research in all areas of chemistry which can benefit from ML and in turn can potentially accelerate the growth of such algorithms. [ABSTRACT FROM AUTHOR]

This work incorporates symbolic regression to propose simple and accurate expressions that fit to material datasets. The incorporation of symbolic regression in physical sciences opens the way to replace "black-box" machine learning techniques with representations that carry the physical meaning and can reveal the underlying mechanism in a purely data-driven approach. The application here is the extraction of analytical equations for the self-diffusion coefficient of the Lennard-Jones fluid by exploiting widely incorporating data from the literature. We propose symbolic formulas of low complexity and error that achieve better or comparable results to well-known microscopic and empirical expressions. Results refer to the material state space both as a whole and in distinct gas, liquid, and supercritical regions. [ABSTRACT FROM AUTHOR]

Reinforcement learning (RL) approaches that combine a tree search with deep learning have found remarkable success in searching exorbitantly large, albeit discrete action spaces, as in chess, Shogi and Go. Many real-world materials discovery and design applications, however, involve multi-dimensional search problems and learning domains that have continuous action spaces. Exploring high-dimensional potential energy models of materials is an example. Traditionally, these searches are time consuming (often several years for a single bulk system) and driven by human intuition and/or expertise and more recently by global/local optimization searches that have issues with convergence and/or do not scale well with the search dimensionality. Here, in a departure from discrete action and other gradient-based approaches, we introduce a RL strategy based on decision trees that incorporates modified rewards for improved exploration, efficient sampling during playouts and a "window scaling scheme" for enhanced exploitation, to enable efficient and scalable search for continuous action space problems. Using high-dimensional artificial landscapes and control RL problems, we successfully benchmark our approach against popular global optimization schemes and state of the art policy gradient methods, respectively. We demonstrate its efficacy to parameterize potential models (physics based and high-dimensional neural networks) for 54 different elemental systems across the periodic table as well as alloys. We analyze error trends across different elements in the latent space and trace their origin to elemental structural diversity and the smoothness of the element energy surface. Broadly, our RL strategy will be applicable to many other physical science problems involving search over continuous action spaces. Reinforcement learning algorithms are emerging as powerful machine learning approaches. This paper introduces a novel machine-learning approach for learning in continuous action space and applies this strategy to the generation of high dimensional potential models for a wide variety of materials. [ABSTRACT FROM AUTHOR]

• We propose a novel procedure for publication subject hierarchical categorization based on the repetition and absence of relevant terms in association rules defined among sets of individual topics. • Our topics are extracted from the category list provided by the SCImago Journal Rank dataset in combination with the journal prior classification provided in the Scopus database to construct a data-driven hierarchy of publication categories. • The procedure can suggest new potential fields, corresponding to sets of publication topics strongly related among them that cannot be suited to any existing field. • The procedure also reveals outliers that are excluded from emerging rules and, not fitting in any other field that should lead to the creation of their own subfield. • Our proposal has been validated using the Jensen–Shannon divergence and supervised machine learning techniques, which relate the categories of a publication and its fields in the hierarchy. Our categorization procedure outperforms the results of the prior preassigned hierarchy. Subject categorization of scientific publications, i.e., journals, book series or conference proceedings, has become a main concern in academia, as publication impact and ranking are considered a basic criterion to evaluate paper quality. Publishers usually propose their own categorization, but they often include only their own publications and their categories might not be coherent with other proposals. Also, due to the dynamic nature of science, new categories may frequently appear. As traditional mechanisms for categorization have been questioned by many authors, a new research line has emerged to improve the category assignment process. Approaches usually rely on assessing publication similarity in terms of topics, co-citation, editorial boards, and/or shared author profiles. In this work, we propose a novel procedure for scientific publication hierarchical categorization based on the repetition or absence of relevant descriptors in association rules among publications. The key idea is that publication categories can be automatically defined by strong associations of nuclear topics. Also, some very specific subcategories can be defined by exclusion from any set of rules. This process can be used to construct a data-driven hierarchy of scientific publication categories from scratch or to improve any existing categorization by discovering new fields. In this paper the proposed algorithm uses SJR descriptors all journals in the SCImago dataset and the three-level classification in the Scopus dataset (covering only 35 % of publications of the SCImago dataset) to discover new categories and assign every journal to the resulting enhanced hierarchy one. We have focused on the field of "Physical Sciences and Engineering", using the SCImago and Scopus datasets from 2019 (30,883 scientific publications). Our procedure combines data engineering techniques with association rules and generates as a result potential new categories and outlier subcategories. To evaluate the suitability of our proposal, we have analyzed classification results based on the original category list and our extended two-level categorization via the Jensen–Shannon divergence and supervised machine-learning techniques. Results reveal the consistency and suitability of our categorization procedure. [ABSTRACT FROM AUTHOR]

Analysing historical patterns of artificial intelligence (AI) adoption can inform decisions about AI capability uplift, but research to date has provided a limited view of AI adoption across different fields of research. In this study we examine worldwide adoption of AI technology within 333 fields of research during 1960–2021. We do this by using bibliometric analysis with 137 million peer-reviewed publications captured in The Lens database. We define AI using a list of 214 phrases developed by expert working groups at the Organisation for Economic Cooperation and Development (OECD). We found that 3.1 million of the 137 million peer-reviewed research publications during the entire period were AI-related, with a surge in AI adoption across practically all research fields (physical science, natural science, life science, social science and the arts and humanities) in recent years. The diffusion of AI beyond computer science was early, rapid and widespread. In 1960 14% of 333 research fields were related to AI (many in computer science), but this increased to cover over half of all research fields by 1972, over 80% by 1986 and over 98% in current times. We note AI has experienced boom-bust cycles historically; the AI "springs" and "winters". We conclude that the context of the current surge appears different, and that interdisciplinary AI application is likely to be sustained. • Share of research publishing on AI rose from 1.2% to 5.3% over last two decades. • Rates of artificial intelligence (AI) adoption higher now than ever before. • Not likely to see a third AI "winter" due to amount and breadth of use. • Impacts of AI range from productivity improvements to paradigm change. [ABSTRACT FROM AUTHOR]

Many measurements in the physical sciences can be cast as counting experiments, where the number of occurrences of a physical phenomenon informs the prevalence of the phenomenon's source. Often, detection of the physical phenomenon (termed signal) is difficult to distinguish from naturally occurring phenomena (termed background). In this case, the discrimination of signal events from background can be performed using classifiers, and they may range from simple, threshold-based classifiers to sophisticated neural networks. These classifiers are often trained and validated to obtain optimal accuracy, however we show that the optimal accuracy classifier does not generally coincide with a classifier that provides the lowest detection limit, nor the lowest quantification uncertainty. We present a derivation of the detection limit and quantification uncertainty in the classifier-based counting experiment case. We also present a novel abstention mechanism to minimize the detection limit or quantification uncertainty a posteriori. We illustrate the method on two data sets from the physical sciences, discriminating Ar-37 and Ar-39 radioactive decay from non-radioactive events in a gas proportional counter, and discriminating neutrons from photons in an inorganic scintillator and report results therefrom. [ABSTRACT FROM AUTHOR]