When using machine learning to model environmental systems, it is often a model’s ability to predict extreme behaviors that yields the highest practical value to policy makers. However, most existing error metrics used to evaluate the performance of environmental machine learning models weigh error equally across test data. Thus, routine performance is prioritized over a model’s ability to robustly quantify extreme behaviors. In this work, we present a new error metric, termed Reflective Error, which quantifies the degree at which our model error is distributed around our extremes, in contrast to existing model evaluation methods that aggregate error over all events. The suitability of our proposed metric is demonstrated on a real-world hydrological modeling problem, where extreme values are of particular concern.

The oriental armyworm, Mythimna separata (Walker), is a highly migratory pest known for its sudden larval outbreaks, which result in severe crop losses. These unpredictable surges pose significant challenges for timely and accurate monitoring, as conventional methods are labour-intensive and prone to errors. To address these limitations, this study investigates the use of machine learning for automated and precise identification of M. separata larval instars. A total of 1577 larval images representing different instar were analysed for geometric, colour, and texture features. Additionally, larval weight was predicted using 13 regression models. Instar identification was conducted using Support Vector Classifier (SVC), Random Forest, and Multi-Layer Perceptron. Key feature contributing to classification accuracy were subsequently identified through permutation feature importance analysis. The results demonstrated the potential of machine learning for automating instar identification with high efficiency and accuracy. Predicted larval weight emerged as a key feature, significantly enhancing the performance of all identification models. Among the tested approaches, BaggingRegressor exhibited the best performance for larval weight prediction (R2 = 98.20%, RMSE = 0.2313), while SVC achieved the highest instar identification accuracy (94%). Overall, the integration of larval weight with other image-derived features proved to be a highly effective strategy. This study demonstrates the efficacy of machine learning in enhancing pest monitoring systems by providing a scalable and reliable framework for precise pest management. The proposed methodology significantly improves larval instar identification accuracy and efficiency, offering actionable insights for implementing targeted biological and chemical control strategies.

In this study, we tackle the challenge of inferring the initial conditions of a Rayleigh–Taylor mixing zone for modelling purposes by analysing zero-dimensional (0-D) turbulent quantities measured at an unspecified time. This approach assesses the extent to which 0-D observations retain the memory of the flow, evaluating their effectiveness in determining initial conditions and, consequently, in predicting the flow’s evolution. To this end, we generated a comprehensive dataset of direct numerical simulations, focusing on miscible fluids with low density contrasts. The initial interface deformations in these simulations are characterised by an annular spectrum parametrised by four non-dimensional numbers. To study the sensitivity of 0-D turbulent quantities to initial perturbation distributions, we developed a surrogate model using a physics-informed neural network (PINN). This model enables computation of the Sobol indices for the turbulent quantities, disentangling the effects of the initial parameters on the growth of the mixing layer. Within a Bayesian framework, we employ a Markov chain Monte Carlo (MCMC) method to determine the posterior distributions of initial conditions and time, given various state variables. This analysis sheds light on inertial and diffusive trajectories, as well as the progressive loss of initial conditions memory during the transition to turbulence. Furthermore, it identifies which turbulent quantities serve as better predictors of Rayleigh–Taylor mixing zone dynamics by more effectively retaining the memory of the flow. By inferring initial conditions and forward propagating the maximum a posteriori (MAP) estimate, we propose a strategy for modelling the Rayleigh–Taylor transition to turbulence.

One of the most significant challenges in research related to nutritional epidemiology is the achievement of high accuracy and validity of dietary data to establish an adequate link between dietary exposure and health outcomes. Recently, the emergence of artificial intelligence (AI) in various fields has filled this gap with advanced statistical models and techniques for nutrient and food analysis. We aimed to systematically review available evidence regarding the validity and accuracy of AI-based dietary intake assessment methods (AI-DIA). In accordance with PRISMA guidelines, an exhaustive search of the EMBASE, PubMed, Scopus and Web of Science databases was conducted to identify relevant publications from their inception to 1 December 2024. Thirteen studies that met the inclusion criteria were included in this analysis. Of the studies identified, 61·5 % were conducted in preclinical settings. Likewise, 46·2 % used AI techniques based on deep learning and 15·3 % on machine learning. Correlation coefficients of over 0·7 were reported in six articles concerning the estimation of calories between the AI and traditional assessment methods. Similarly, six studies obtained a correlation above 0·7 for macronutrients. In the case of micronutrients, four studies achieved the correlation mentioned above. A moderate risk of bias was observed in 61·5 % (n 8) of the articles analysed, with confounding bias being the most frequently observed. AI-DIA methods are promising, reliable and valid alternatives for nutrient and food estimations. However, more research comparing different populations is needed, as well as larger sample sizes, to ensure the validity of the experimental designs.

Data-based methods have gained increasing importance in engineering. Success stories are prevalent in areas such as data-driven modeling, control, and automation, as well as surrogate modeling for accelerated simulation. Beyond engineering, generative and large-language models are increasingly helping with tasks that, previously, were solely associated with creative human processes. Thus, it seems timely to seek artificial-intelligence-support for engineering design tasks to automate, help with, or accelerate purpose-built designs of engineering systems for instance in mechanics and dynamics, where design so far requires a lot of specialized knowledge. Compared with established, predominantly first-principles-based methods, the datasets used for training, validation, and test become an almost inherent part of the overall methodology. Thus, data publishing becomes just as important in (data-driven) engineering science as appropriate descriptions of conventional methodology in publications in the past. However, in mechanics and dynamics, quite widely, still traditional publishing practices are prevalent that largely do not yet take into account the rising role of data as much as that may already be the case in pure data-scientific research. This article analyzes the value and challenges of data publishing in mechanics and dynamics, in particular regarding engineering design tasks, showing that the latter raise also challenges and considerations not typical in fields where data-driven methods have been booming originally. Researchers currently find barely any guidance to overcome these challenges. Thus, ways to deal with these challenges are discussed and a set of examples from across different design problems shows how data publishing can be put into practice.

Machine learning has exhibited substantial success in the field of natural language processing (NLP). For example, large language models have empirically proven to be capable of producing text of high complexity and cohesion. However, at the same time, they are prone to inaccuracies and hallucinations. As these systems are increasingly integrated into real-world applications, ensuring their safety and reliability becomes a primary concern. There are safety critical contexts where such models must be robust to variability or attack and give guarantees over their output. Computer vision had pioneered the use of formal verification of neural networks for such scenarios and developed common verification standards and pipelines, leveraging precise formal reasoning about geometric properties of data manifolds. In contrast, NLP verification methods have only recently appeared in the literature. While presenting sophisticated algorithms in their own right, these papers have not yet crystallised into a common methodology. They are often light on the pragmatical issues of NLP verification, and the area remains fragmented. In this paper, we attempt to distil and evaluate general components of an NLP verification pipeline that emerges from the progress in the field to date. Our contributions are twofold. First, we propose a general methodology to analyse the effect of the embedding gap – a problem that refers to the discrepancy between verification of geometric subspaces, and the semantic meaning of sentences which the geometric subspaces are supposed to represent. We propose a number of practical NLP methods that can help to quantify the effects of the embedding gap. Second, we give a general method for training and verification of neural networks that leverages a more precise geometric estimation of semantic similarity of sentences in the embedding space and helps to overcome the effects of the embedding gap in practice.

The greatest challenge in pressure reconstruction from the measured velocity fields is that the error of material acceleration is significantly contaminated due to error propagation. Particularly for flows with moving boundaries, accurate boundary velocities are difficult to obtain due to error propagation, and a complex boundary processing technique is needed to treat the moving boundaries. The present work proposes a machine-learning-based method to determine the pressure for incompressible flows with moving boundaries. The proposed network consists of two neural networks: one network, named the boundary network, is used to track the Lagrangian boundary points; the other physics-informed neural network, named the flow network, is adopted to approximate the flow fields. These two networks are coupled by imposing boundary conditions. We further propose a new dynamic weight strategy for the loss terms to guarantee convergence and stability. The performance of the proposed method is validated by two examples: the flow over an oscillating cylinder and the flow around a swimming fish. The proposed method can accurately determine the pressure fields and boundary motion from synthetic particle image velocimetry (PIV) flow fields. Moreover, this method can also predict the boundary and pressure at a given instant without supervised data. Finally, this method was applied to reconstruct the pressure from the two-dimensional and three-dimensional PIV velocities of the left ventricle. All of the results indicate that the proposed method can accurately reconstruct the pressure fields for flows with moving boundaries and is a novel method for surface pressure estimation.

Deep neural networks and other modern machine learning models are often susceptible to adversarial attacks. Indeed, an adversary may often be able to change a model’s prediction through a small, directed perturbation of the model’s input – an issue in safety-critical applications. Adversarially robust machine learning is usually based on a minmax optimisation problem that minimises the machine learning loss under maximisation-based adversarial attacks. In this work, we study adversaries that determine their attack using a Bayesian statistical approach rather than maximisation. The resulting Bayesian adversarial robustness problem is a relaxation of the usual minmax problem. To solve this problem, we propose Abram – a continuous-time particle system that shall approximate the gradient flow corresponding to the underlying learning problem. We show that Abram approximates a McKean–Vlasov process and justify the use of Abram by giving assumptions under which the McKean–Vlasov process finds the minimiser of the Bayesian adversarial robustness problem. We discuss two ways to discretise Abram and show its suitability in benchmark adversarial deep learning experiments.

During the coronavirus pandemic in the United Kingdom, media outlets shifted their focus from divisive political issues to more neutral topics like lifestyle, sports, and entertainment. This study explores how this change in media content relates to partisan divides in satisfaction with democracy. Using data from a representative survey of 201,144 individuals, we linked respondents’ perceptions of democratic performance to their daily media exposure. We did so by analysing 1.5 million tweets from British newspapers using a topic modelling algorithm to identify shifts in topic salience and sentiment using sentiment analysis. Our findings reveal a decline in partisan media exposure during the pandemic, associated with increased satisfaction with democracy at both individual and collective levels, and a narrowing of cross-party divides. These results contribute to discussions on affective polarization, the winner-loser gap in democratic evaluation, and media framing effects, highlighting the potential influence of depoliticized news coverage on democratic attitudes.

With the increasing volume of scientific literature, there is a need to streamline the screening process for titles and abstracts in systematic reviews, reduce the workload for reviewers, and minimize errors. This study validated artificial intelligence (AI) tools, specifically Llama 3 70B via Groq’s application programming interface (API) and ChatGPT-4o mini via OpenAI’s API, for automating this process in biomedical research. It compared these AI tools with human reviewers using 1,081 articles after duplicate removal. Each AI model was tested in three configurations to assess sensitivity, specificity, predictive values, and likelihood ratios. The Llama 3 model’s LLA_2 configuration achieved 77.5% sensitivity and 91.4% specificity, with 90.2% accuracy, a positive predictive value (PPV) of 44.3%, and a negative predictive value (NPV) of 97.9%. The ChatGPT-4o mini model’s CHAT_2 configuration showed 56.2% sensitivity, 95.1% specificity, 92.0% accuracy, a PPV of 50.6%, and an NPV of 96.1%. Both models demonstrated strong specificity, with CHAT_2 having higher overall accuracy. Despite these promising results, manual validation remains necessary to address false positives and negatives, ensuring that no important studies are overlooked. This study suggests that AI can significantly enhance efficiency and accuracy in systematic reviews, potentially revolutionizing not only biomedical research but also other fields requiring extensive literature reviews.

The 2018/2019 trade conflict between the United States and China impacted a broad array of agricultural products, including soybeans. Previous trade studies using gravity models fail to account for trends and complex seasonal patterns observed in the data. This study uses a machine learning (ML) approach to estimate losses in soybean export value and volume from the trade war. We find that models using ML techniques outperform traditional models and estimate losses in the value of soybean exports of $10.16 billion/year. The ML models fit the complex export trade data series well, highlighting the importance of utilizing improved modeling approaches.

The study aimed to delve into the incidence and risk factors associated with myocarditis and pericarditis following SARS-COV-2-19 vaccination, addressing a notable gap in understanding the safety profile of vaccinations. Through meticulous data selection from the National Health Insurance System (NHIS) database of Korea, the researchers employed both a case-crossover study and a nested case-control design to analyze temporal patterns and risk factors related to carditis occurrences post-immunization. Key findings revealed a significant association between SARS-COV-2-19 vaccination and the occurrence of carditis, with a strong temporal correlation observed within 10 days post-vaccination. Noteworthy factors contributing to carditis risk included the duration between vaccination and carditis, specific comorbidities and medication use. The study concluded by recommending an extended post-vaccination surveillance duration of at least 10 days and underscored the importance of considering individual medical histories and concurrent medication use in assessing vaccine-induced carditis risk. This study might contribute to understanding vaccine safety profiles and emphasizes the significance of comprehensive post-vaccination monitoring protocols.

The present study explores the value of machine learning techniques in the classification of communication content in experiments. Previously human-coded datasets are used to both train and test algorithm-generated models that relate word counts to categories. For various games, the computer models of the classification are able to match out-of-sample the human classification to a considerable extent. The analysis raises hope that the substantial effort going into such studies can be reduced by using computer algorithms for classification. This would enable a quick and replicable analysis of large-scale datasets at reasonable costs and widen the applicability of such approaches. The paper gives an easily accessible technical introduction into the computational method.