Older adults often show memory deficits, but these deficits can be reduced when newly learned information is consistent with one's schemas (prior knowledge). For example, research has found similar memory performance for young and older adults when remembering realistic market-value grocery items and prices; however, age-related differences are more prominent for overpriced items, which are inconsistent with schemas. In the present work, we examined how labelling items as free may impact memory for prices, and if curiosity may influence how younger and older adults remember price information. Experiment 1 investigated young and older adults' memory for free, market-priced, and overpriced items. In Experiment 2, participants' curiosity for learning the items and their prices was investigated to see if curiosity can be predictive of performance accuracy for information that is consistent and inconsistent with their schemas. In Experiment 1, participants were shown grocery store items and were tested on the exact prices of these items and the categories they belong to. In Experiment 2, participants were first shown items and asked how curious they were to learn the price of the item. Participants were then shown the grocery item's price and category label, and were later tested on the associations of these items. Across both experiments, older adults were more accurate in recalling market-priced and free items compared to overpriced items. In contrast, younger adults did not show significant differences across price conditions. In Experiment 2, state curiosity significantly predicted accuracy in recalling exact prices, with this relationship being particularly strong among older adults. Findings suggest that older adults benefit more from schematic support when remembering price information and that state curiosity enhances memory accuracy, especially for older adults. Schematic support and state curiosity may help mitigate age-related memory deficits.

Due to the complexity of assembly processes in complex products, even slight deviations can result in quality problems. Quality problems often stem from variations in quality characteristics that, when propagated and superimposed through the assembly flow, exceed acceptable thresholds. Modelling these causal effects at the quality characteristic level and establishing a causal network is an effective strategy for quality control. This paper proposes a data- and knowledge-driven approach for causal modelling and quality control in complex product assembly. First, the propagation paths of quality characteristic variations are extracted as prior knowledge and incorporated into a reinforcement learning algorithm to guide causal graph construction and improve the accuracy of causal modelling. Second, a quality control framework integrating root cause analysis and prediction is developed based on the established causal network. A Bayesian method is applied to provide probabilistic guidance for root cause analysis, while the causal network is used to identify and eliminate characteristics unrelated to the target characteristic, thereby enhancing the accuracy of quality prediction. Finally, the proposed method is validated using an aircraft assembly case study. Experimental results demonstrate its feasibility and effectiveness in enhancing quality control in complex product assembly.

Synesthesia is a neurological phenomenon in which healthy individuals experience additional, automatic, and consistent perceptions unrelated to veridical sensory input. For most (but not all) synesthetes, this additional experience is a color: for example, grapheme-color synesthetes experience colors for letters of the alphabet. Measuring these color associations is of central importance to synesthesia research, but there is no standard color picker "tool" that researchers can adapt to use in their own experiments: each researcher must code their own. This is a barrier to entry for synesthesia research, and additionally creates potential methodological confounds because different researchers make color pickers with different properties. SynesthesiaColorPicker is an open-source, mobile-friendly color picker tool that can be integrated with two popular online experiment platforms (Qualtrics and lab.js/Open Lab) without any prior programming knowledge. The templates, underlying JavaScript code, and detailed instructions are available for download on a GitHub repository. Furthermore, a comparison between data collected with SynesthesiaColorPicker and with the Synesthesia Battery shows that two methodological design choices in SynesthesiaColorPicker overcome measurable confounds in existing color picker methodology.

Abstract To solve the limitation of traditional reliability evaluation methods for electromechanical equipment in dealing with the uncertainty and dynamic evolution characteristics of operating data, a dynamic reliability evaluation model integrating physical degradation mechanism and data-driven is constructed in this study. Based on the Wiener process, this model constructs a physical model that can describe the random degradation of equipment performance with time to represent the internal random failure process of equipment. Furthermore, Bayesian inference framework is introduced, and Markov Chain Monte Carlo (MCMC) algorithm is used to integrate the real-time monitoring data into the model to realize the dynamic online updating and uncertainty quantification of the parameters of the degraded model. By modifying the prior knowledge with real-time data, the probability distribution of the Remaining Useful Life (RUL) of the equipment at any time can be obtained, and its reliability can be dynamically evaluated. To verify the effectiveness and superiority of the proposed model, this study uses the degradation dataset of turbofan engine for example analysis. The experimental results show that the proposed model is superior in predicting RUL. Its Root Mean Square Error (RMSE) is as low as 12.88, which is significantly better than the benchmark models such as Long Short-Term Memory network (LSTM). The model can also effectively quantify the prediction uncertainty, and its Prediction Interval Coverage Probability (PICP) reaches 94.1%. The research proves that the fusion model can effectively integrate the physical prior information with the actual monitoring data, and realize a more accurate and dynamic reliability evaluation of electromechanical equipment.

This study examines the alignment between whale-watching experiences and tourist expectations in three different destinations. Whale-watching is a global tourist activity, with locations such as the Canary Islands (Spain) and the Azores (Portugal) in Macaronesia rapidly becoming prime spots for these marine activities. Those locations attract a significant number of tourists with varying recreational interests and diverse perceptions of each destination and its natural resources, including marine wildlife megafauna species that can be seen. While often marketed as sustainable tourism, the ecological impacts of whale-watching are a matter of concern. Evolving whale-watching practices may reinforce or diminish the effectiveness of conservation and environmental education efforts. In this regard, exploring whale watchers' expectatives, preferences, previous experiences, level of satisfaction, and environmental information may help to assess better practices and sustainable tourism initiatives. This study employed a multidisciplinary approach, incorporating a series of questionnaires that explored whale watchers' expectations and overall satisfaction before and after sea trips in the Tenerife and El Hierro Islands (Canary Islands, Spain), as well as in São Miguel (Azores, Portugal). The findings highlight differences across three study cases: El Hierro attracted more experienced and oriented tourists, while Tenerife and São Miguel received more generalist visitors. Satisfaction was closely linked to the number of cetacean sightings. Many participants who did not mention specific species expressed open preferences such as wanting to "see everything" or "whatever is possible", reflecting limited prior knowledge. The study highlights the importance of tailoring whale-watching strategies to tourist profiles by enhancing communication, adjusting group sizes and vessel types, and reinforcing conservation messaging to ensure both positive experiences and long-term ecological sustainability.

Recently, backdoor attacks on Deep Neural Networks (DNNs) have raised urgent security threats, which can manipulate the behavior of an attacked model by embedding the backdoor trigger into the input. Since triggers can be designed to be stealthy and hard to recognize by the naked eye, segmenting these triggers in backdoor samples becomes a significant challenge. However, finding triggers embedded by the attacker can be crucial for analyzing the attacks and formulating a defense strategy. Therefore, in this paper, we propose the Backdoor Trigger Segmentation (BTS) task with a comprehensive benchmark consisting of 8 attack methods, 8 unique triggers, and 179 attack settings for image or text data. Moreover, we construct a mathematical system for BTS, abstracting various backdoor triggers into a unified theoretical framework. Based on the theoretical guarantees, we propose a unified Trigger Locator (TriLoc) algorithm to segment various triggers in backdoor samples of both image and text modalities, without prior knowledge of triggers. Extensive experimental results on our benchmark demonstrate the superior performance of our algorithm compared to state-of-the-art methods. Our benchmark and code are available at https://github.com/LivXue/Backdoor-Trigger-Segmentation.

Sample size determination (SSD) is essential in statistical inference and hypothesis testing, as it directly affects the accuracy and power of the analysis. We propose a SSD methodology for one and two-sample t-tests that ensures clinical relevance using a pre-determined unstandardized effect size. Our novel approach leverages Bayesian meta-analysis to account for the uncertainty surrounding the variance, a common issue in SSD. By incorporating prior knowledge from related studies via a Bayesian gamma-inverse gamma model, we obtain an informative posterior predictive distribution for the variance that leads to better decisions about sample size. For efficient posterior sampling, we propose an empirical Bayes approach which is implemented with systematic sampling to facilitate computation. Simulations and empirical studies demonstrate that our methodology outperforms other aggregate approaches (simple average, weighted average, median) in variance estimation for SSD, especially in meta-analyses with large disparity in sample size and moderate variance. Thus, it offers a robust and practical solution for sample size determination in t-tests.

Domain adaptation for low-dose CT denoising via pretraining and self-supervised fine-tuning.

LMcast: A pretrained language model guided long-term memory transformer for precipitation nowcasting.

Real-life evidence on point-of-care HIV Ag/Ab rapid testing in Türkiye: Early diagnosis and linkage to care.

A functional system-informed graph neural network framework to quantify interpretable brain dysfunction in ASD.

GNNRL-smoothing: A prior-free reinforcement learning model for mesh optimization.

Regulatory approaches for platform-based vaccine development in Japan: Insights from PMDA's experience with COVID-19 and RSV vaccines.

Bidirectional effects between working memory and math in children with and without math disabilities: Does working memory or math drive the system?

Abstract This research addresses shared intention prediction in multi-human, multi-robot environments. We propose an intention prediction pipeline based on Bayesian inference, enabling robots to predict humans’ navigational intent. Our pipeline uses prior semantic knowledge about potential goal destinations of the humans. We furthermore investigated different strategies for robots to share information to improve their predictions. A dataset was collected specifically for testing our pipeline and comparing different sharing strategies. The dataset consists of camera feeds from two robots observing two humans performing simple pick-up tasks. The pipeline correctly predicts 63% of the cases. Implementing a relatively simple data sharing strategy increased accuracy to 77%. However, a poorly designed data sharing strategy reduced accuracy to 47%. A validation dataset was collected in a more realistic office setting, which supports the observed relative differences in performance, despite the increased complexity of this environment. Our key finding is thus that the choice of sharing strategy can significantly impact prediction performance. The insights gained from our experiments can help progress the development of more effective shared intention prediction in multi-human, multi-robot teams.

Data-driven modeling with prior system knowledge

Early detection of cardiovascular disease and implementation of evidence-based treatments can reduce cardiovascular morbidity and mortality. Medical algorithms and decision-making tools provide a compelling option for screening, risk prediction, and treatment management. Such digital tools have the potential to aid both healthcare professionals and patients, providing support to decrease unwarranted diagnostic and treatment variability while guiding personalized care, with the overall objective of improving clinical outcomes. However, incorporating digital tools in healthcare settings is challenging, and evidence the required to support their adoption and understand the limitations can be lacking. A multinational panel of expert cardiologists and emergency physicians across North America, Europe, and Oceania gathered to deliberate on the current landscape of digital tools and medical algorithms, drawing on prior clinical experiences and knowledge of country-specific regulations. In this viewpoint, the evidence to support and guide the adoption of digital tools in cardiovascular clinical practice and the necessary components for successful integration into clinical workflows were discussed. Digital tools must be developed with the needs of the healthcare professionals, other relevant stakeholders (eg, administration personnel), and patients in mind to give them the best chance of widespread adoption. Academia, industry, and regulatory bodies should work together to cultivate and accelerate the implementation of digital tools in healthcare. The considerations discussed here may help decision makers to determine if a digital tool has the components necessary to integrate into the clinical workflow successfully.

Delayed patient isolation and associated factors during the mpox outbreak in Uganda, July-December 2024.

Continual learning: A systematic literature review.

Enhancing mixture-of-experts model with prior knowledge for infrared and visible image fusion in complex degraded environments