Hasil untuk "Science (General)"

Haotian Liu, Chunyuan Li, Qingyang Wu et al.

Instruction tuning large language models (LLMs) using machine-generated instruction-following data has improved zero-shot capabilities on new tasks, but the idea is less explored in the multimodal field. In this paper, we present the first attempt to use language-only GPT-4 to generate multimodal language-image instruction-following data. By instruction tuning on such generated data, we introduce LLaVA: Large Language and Vision Assistant, an end-to-end trained large multimodal model that connects a vision encoder and LLM for general-purpose visual and language understanding.Our early experiments show that LLaVA demonstrates impressive multimodel chat abilities, sometimes exhibiting the behaviors of multimodal GPT-4 on unseen images/instructions, and yields a 85.1% relative score compared with GPT-4 on a synthetic multimodal instruction-following dataset. When fine-tuned on Science QA, the synergy of LLaVA and GPT-4 achieves a new state-of-the-art accuracy of 92.53%. We make GPT-4 generated visual instruction tuning data, our model and code base publicly available.

J. Stilgoe, R. Owen, P. Macnaghten

The governance of emerging science and innovation is a major challenge for contemporary democracies. In this paper we present a framework for understanding and supporting efforts aimed at ‘responsible innovation’. The framework was developed in part through work with one of the first major research projects in the controversial area of geoengineering, funded by the UK Research Councils. We describe this case study, and how this became a location to articulate and explore four integrated dimensions of responsible innovation: anticipation, reflexivity, inclusion and responsiveness. Although the framework for responsible innovation was designed for use by the UK Research Councils and the scientific communities they support, we argue that it has more general application and relevance.

C. Felser, G. Fecher, B. Balke

A. Kaufmann, M. Gupta

X. Duan, Charles M. Lieber

K. Burnham, David R. Anderson

N. Kirkham, Jonathan A. Slemmer, Scott P. Johnson

S. Lewandowsky, Gilles E. Gignac, K. Oberauer

Background Among American Conservatives, but not Liberals, trust in science has been declining since the 1970's. Climate science has become particularly polarized, with Conservatives being more likely than Liberals to reject the notion that greenhouse gas emissions are warming the globe. Conversely, opposition to genetically-modified (GM) foods and vaccinations is often ascribed to the political Left although reliable data are lacking. There are also growing indications that rejection of science is suffused by conspiracist ideation, that is the general tendency to endorse conspiracy theories including the specific beliefs that inconvenient scientific findings constitute a “hoax.” Methodology/Principal findings We conducted a propensity weighted internet-panel survey of the U.S. population and show that conservatism and free-market worldview strongly predict rejection of climate science, in contrast to their weaker and opposing effects on acceptance of vaccinations. The two worldview variables do not predict opposition to GM. Conspiracist ideation, by contrast, predicts rejection of all three scientific propositions, albeit to greatly varying extents. Greater endorsement of a diverse set of conspiracy theories predicts opposition to GM foods, vaccinations, and climate science. Conclusions Free-market worldviews are an important predictor of the rejection of scientific findings that have potential regulatory implications, such as climate science, but not necessarily of other scientific issues. Conspiracist ideation, by contrast, is associated with the rejection of all scientific propositions tested. We highlight the manifold cognitive reasons why conspiracist ideation would stand in opposition to the scientific method. The involvement of conspiracist ideation in the rejection of science has implications for science communicators.

W. Buytaert, Z. Zulkafli, S. Grainger et al.

The participation of the general public in the research design, data collection and interpretation process together with scientists is often referred to as citizen science. While citizen science itself has existed since the start of scientific practice, developments in sensing technology, data processing and visualisation, and communication of ideas and results, are creating a wide range of new opportunities for public participation in scientific research. This paper reviews the state of citizen science in a hydrological context and explores the potential of citizen science to complement more traditional ways of scientific data collection and knowledge generation for hydrological sciences and water resources management. Although hydrological data collection often involves advanced technology, the advent of robust, cheap and low-maintenance sensing equipment provides unprecedented opportunities for data collection in a citizen science context. These data have a significant potential to create new hydrological knowledge, especially in relation to the characterisation of process heterogeneity, remote regions, and human impacts on the water cycle. However, the nature and quality of data collected in citizen science experiments is potentially very different from those of traditional monitoring networks. This poses challenges in terms of their processing, interpretation, and use, especially with regard to assimilation of traditional knowledge, the quantification of uncertainties, and their role in decision support. It also requires care in designing citizen science projects such that the generated data complement optimally other available knowledge. Lastly, we reflect on the challenges and opportunities in the integration of hydrologically-oriented citizen science in water resources management, the role of scientific knowledge in the decision-making process, and the potential contestation to established community institutions posed by co-generation of new knowledge.

José L. Duarte, Jarret T. Crawford, Charlotta Stern et al.

Zhao-Song Li, Chao Liu, Xiao-Wei Li et al.

Abstract As a frontier technology, holography has important research values in fields such as bio-micrographic imaging, light field modulation and data storage. However, the real-time acquisition of 3D scenes and high-fidelity reconstruction technology has not yet made a breakthrough, which has seriously hindered the development of holography. Here, a novel holographic camera is proposed to solve the above inherent problems completely. The proposed holographic camera consists of the acquisition end and the calculation end. At the acquisition end of the holographic camera, specially configured liquid materials and liquid lens structure based on voice-coil motor-driving are used to produce the liquid camera, so that the liquid camera can quickly capture the focus stack of the real 3D scene within 15 ms. At the calculation end, a new structured focus stack network (FS-Net) is designed for hologram calculation. After training the FS-Net with the focus stack renderer and learnable Zernike phase, it enables hologram calculation within 13 ms. As the first device to achieve real-time incoherent acquisition and high-fidelity holographic reconstruction of a real 3D scene, our proposed holographic camera breaks technical bottlenecks of difficulty in acquiring the real 3D scene, low quality of the holographic reconstructed image, and incorrect defocus blur. The experimental results demonstrate the effectiveness of our holographic camera in the acquisition of focal plane information and hologram calculation of the real 3D scene. The proposed holographic camera opens up a new way for the application of holography in fields such as 3D display, light field modulation, and 3D measurement.

Ali AbuGneam, Somayeh Nemati, Afshin Babaei

In this research, we propose a new numerical method for solving a class of distributed-order fractional partial differential equations, specifically focusing on distributed-order time fractional wave-diffusion equations. The method begins by introducing a novel generalization of Bernoulli wavelets and deriving an exact result for the Riemann–Liouville integral of these new basis functions. Utilizing the Gauss–Legendre quadrature formula and a strategically chosen set of collocation points, along with approximations for the unknown function and its derivatives, we transform the problem into a system of algebraic equations. An error analysis is then conducted for the approximation of a bivariate function using fractional-order Bernoulli wavelets. Finally, three examples are solved to demonstrate the method’s applicability and accuracy, with the numerical results confirming its effectiveness. These findings demonstrate that the parameters of the basis functions can be adjusted to suit the given problem, thereby enhancing the accuracy of the method.

Seungjae Han, Taeyoung Oh, Soohyeon Lee et al.

With continuous advancements in autonomous driving technology, systematic and reliable safety verification is becoming increasingly important. However, despite the active development of various X-in-the-loop simulation (XILS) platforms to validate autonomous driving systems (ADSs), standardized evaluation frameworks for assessing the credibility of the simulation platforms themselves remain lacking. Therefore, we propose a novel integrated credibility-assessment methodology that combines dynamics-based fidelity assessment, parameter-based reliability assessment, and scenario-based reliability assessment. These three techniques evaluate the similarity and consistency between XILS and real-world test data based on statistical and mathematical comparisons. The three consistency measures are then utilized to derive a dynamics-based correlation metric for fidelity, along with parameter-based and scenario-based correlation and applicability metrics for reliability. The novel contribution of this paper lies in a geometric similarity analysis methodology that significantly enhances the efficiency of credibility assessment. We propose a methodology that enables geometric similarity assessment through spider chart visualization of metrics derived from the credibility-assessment process and shape comparison, based on Procrustes, Fréchet, and Hausdorff distances. As a result, speed is not a dominant factor for credibility evaluation, enabling assessment with a single representative speed test; the framework simplifies the XILS evaluation and enhances ADS validation efficiency.

Ifiok Udoidiok, Fuhao Li, Jielun Zhang

Machine learning (ML) has become a cornerstone of critical applications, but its vulnerability to data poisoning attacks threatens system reliability and trustworthiness. Prior studies have begun to investigate the impact of data poisoning and proposed various defense or evaluation methods; however, most efforts remain limited to quantifying performance degradation, with little systematic comparison of internal behaviors across model architectures under attack and insufficient attention to interpretability for revealing model vulnerabilities. To tackle this issue, we build a reproducible evaluation pipeline and emphasize the importance of integrating robustness with interpretability in the design of secure and trustworthy ML systems. To be specific, we propose a unified poisoning evaluation framework that systematically compares traditional ML models, deep neural networks, and large language models under three representative attack strategies including label flipping, random corruption, and adversarial insertion, at escalating severity levels of 30%, 50%, and 75%, and integrate LIME-based explanations to trace the evolution of model reasoning. Experimental results demonstrate that traditional models collapse rapidly under label noise, whereas Bayesian LSTM hybrids and large language models maintain stronger resilience. Further interpretability analysis uncovers attribution failure patterns, such as over-reliance on neutral tokens or misinterpretation of adversarial cues, providing insights beyond accuracy metrics.

F. Grin, L. Marácz, N. Pokorn

Publisher Summary This chapter attracts the attention of clay scientists in academe and industry as well as in politics (as research needs funding), and focuses on the importance of clay science to society and the quality of life. The economic benefits seem evident because clays are abundant, widespread, and inexpensive compared with other raw materials. The chapter discusses the industrial and environmental importance of clays and clay minerals. The great variety of physical, chemical, and thermal treatments that may be used to modify clays and clay minerals provide unlimited scope for future applications, particularly in terms of protecting the environment. Because of the multidisciplinary nature of clay science, its teaching is another challenging task. By learning about the mineralogical, physico-chemical, and industrial aspects of clay science, students would not only gain an appreciation of the “scientific method” and the physical environment but also find suitable employment and a fulfilling career.

Limin Chuan, Jingjuan Zhao, Shijie Qi et al.

From the perspective of project and paper datasets, research frontier recognition in the field of agricultural resources and the environment using the Latent Dirichlet Allocation (LDA) topic extraction model was studied. By combining the wisdom of domain experts to judge the similarities and differences of clustering topics between the two data sources, multidimensional indicators, such as the emerging degree, attention degree, innovation degree, and intersection degree, were comprehensively constructed for frontier identification. The methods for hot research frontiers, emerging research frontiers, extinction research frontiers, and potential research frontiers were proposed. The empirical research in the field of agricultural resources and the environment showed that the “interaction mechanism of plant–rhizosphere–microbial diversity” was a hot research frontier in the years 2016–2021. The themes of “wastewater treatment technology and efficient utilization of water resources”, the “value-added utilization of agricultural wastes and sustainable development”, the “soil ecological response mechanism under agronomic management measures”, and the “mechanism of soil landslide, erosion, degradation and prediction evaluation” were judged as potential research frontiers. The theme of “ecosystems management and pollution control of agricultural and animal husbandry” was recognized as an emerging research frontier. The results confirm that the fusion method of extracting topics from project and paper data, combined with expert intelligence and frontier indicators for fine classification of frontiers, is an optional approach. This study provides strong support for accurately identifying the forefront of scientific research, grasping the latest research progress, efficiently allocating scientific and technological resources, and promoting technological innovation.

Yao He, Jing Yang, Shaobo Li et al.

Abstract Catastrophic forgetting in neural networks is a common problem, in which neural networks lose information from previous tasks after training on new tasks. Although adopting a regularization method that preferentially retains the parameters important to the previous task to avoid catastrophic forgetting has a positive effect; existing regularization methods cause the gradient to be near zero because the loss is at the local minimum. To solve this problem, we propose a new continuous learning method with Bayesian parameter updating and weight memory (CL-BPUWM). First, a parameter updating method based on the Bayes criterion is proposed to allow the neural network to gradually obtain new knowledge. The diagonal of the Fisher information matrix is then introduced to significantly minimize computation and increase parameter updating efficiency. Second, we suggest calculating the importance weight by observing how changes in each network parameter affect the model prediction output. In the process of model parameter updating, the Fisher information matrix and the sensitivity of the network are used as the quadratic penalty terms of the loss function. Finally, we apply dropout regularization to reduce model overfitting during training and to improve model generalizability. CL-BPUWM performs very well in continuous learning for classification tasks on CIFAR-100 dataset, CIFAR-10 dataset, and MNIST dataset. On CIFAR-100 dataset, it is 0.8%, 1.03% and 0.75% higher than the best performing regularization method (EWC) in three task partitions. On CIFAR-10 dataset, it is 2.25% higher than the regularization method (EWC) and 0.7% higher than the scaled method (GR). It is 0.66% higher than the regularization method (EWC) on the MNIST dataset. When the CL-BPUWM method was combined with the brain-inspired replay model under the CIFAR-100 and CIFAR-10 datasets, the classification accuracy was 2.35% and 5.38% higher than that of the baseline method, BI-R + SI.

Felix E Y Aggor, Martinna Bertolini, Bianca M Coleman et al.

Oropharyngeal candidiasis (OPC) is the most common human fungal infection, arising typically from T cell immune impairments. IL-17 and IL-22 contribute individually to OPC responses, but here we demonstrate that the combined actions of both cytokines are essential for resistance to OPC. Mice lacking IL-17RA and IL-22RA1 exhibited high fungal loads in esophagus- and intestinal tract, severe weight loss, and symptoms of colitis. Ultimately, mice succumbed to infection. Dual loss of IL-17RA and IL-22RA impaired expression of small proline rich proteins (SPRRs), a class of antimicrobial effectors not previously linked to fungal immunity. Sprr2a1 exhibited direct candidacidal activity in vitro, and Sprr1-3a-/- mice were susceptible to OPC. Thus, cooperative actions of Type 17 cytokines mediate oral mucosal anti-Candida defenses and reveal a role for SPRRs.

Ashagrie Sharew Iyassu, Haile Mekonnen Fenta, Zelalem G. Dessie et al.

Abstract Background In causal analyses, some third factor may distort the relationship between the exposure and the outcome variables under study, which gives spurious results. In this case, treatment groups and control groups that receive and do not receive the exposure are different from one another in some other essential variables, called confounders. Method Place of birth was used as exposure variable and age-specific childhood vaccination status was used as outcome variables. Three approaches of confounder selection techniques such as all pre-treatment covariates, outcome cause covariates, and common cause covariates were proposed. Multiple logistic regression was used to estimate the propensity score for inverse probability treatment weighting (IPTW) confounder adjustment techniques. The proportional odds model was used to estimate the causal effect of place of birth on age-specific childhood vaccination. To validate the result obtained from observed data, we used a plasmode simulation of resampling 1000 samples from actual data 500 times. Result Outcome cause and common cause confounder identification techniques gave comparable results in terms of treatment effect in the plasmode data. However, outcome causes that contain common causes and predictors of the outcome confounder identification gave relatively better treatment effect results. The treatment effect result in the IPTW confounder adjustment method was better than that of the regression adjustment method. The effect of place of birth on log odds of cumulative probability of age-specific childhood vaccination was 0.36 with odds ratio of 1.43 for higher level vaccination status. Conclusion It is essential to use plasmode simulation data to validate the reproducibility of the proposed methods on the observed data. It is important to use outcome-cause covariates to adjust their confounding effect on the outcome. Using inverse probability treatment weighting gives unbiased treatment effect results as compared to the regression method of confounder adjustment. Institutional delivery increases the likelihood of childhood vaccination at the recommended schedule.

Ho Soon Choi

This study aims to produce renewable energy by applying a solar-energy-harvesting architectural design using solar panels on the facade of a building. To install as many solar panels as possible on the building elevation, the Signal Box auf dem Wolf, located in Basel, Switzerland, was selected as the research target. The solar panels to be installed on the facade of the Signal Box auf dem Wolf are planned such that they are able to move according to the optimal tilt angle every month to allow maximal energy generation. The kinetic photovoltaic facade system and the simulation of renewable energy generation were implemented using a parametric design. The novelty of this study is the development of a kinetic photovoltaic facade system using a parametric design algorithm. From the perspective of renewable energy in the field of architecture, the kinetic photovoltaic facade system developed in this study has the advantage of producing maximal renewable energy according to the optimal tilt angle of the solar panels. Additionally, building facades that move according to the optimal tilt angle will contribute to the expansion of the field of sustainable architectural design.