10784 research outputs found
Sort by
Addressing the advantages and limitations of using Aethalometer data to determine the optimal absorption Ångström exponents (AAEs) values for eBC source apportionment
No full textPublisher Copyright: © 2025 The AuthorsThe apportionment of equivalent black carbon (eBC) to combustion sources from liquid fuels (mainly fossil; eBCLF) and solid fuels (mainly non-fossil; eBCSF) is commonly performed using data from Aethalometer instruments (AE approach). This study evaluates the feasibility of using AE data to determine the absorption Ångström exponents (AAEs) for liquid fuels (AAELF) and solid fuels (AAESF), which are fundamental parameters in the AE approach. AAEs were derived from Aethalometer data as the fit in a logarithmic space of the six absorption coefficients (470–950 nm) versus the corresponding wavelengths. The findings indicate that AAELF can be robustly determined as the 1st percentile (PC1) of AAE values from fits with R2 > 0.99. This R2-filtering was necessary to remove extremely low and noisy-driven AAE values commonly observed under clean atmospheric conditions (i.e., low absorption coefficients). Conversely, AAESF can be obtained from the 99th percentile (PC99) of unfiltered AAE values. To optimize the signal from solid fuel sources, winter data should be used to calculate PC99, whereas summer data should be employed for calculating PC1 to maximize the signal from liquid fuel sources. The derived PC1 (AAELF) and PC99 (AAESF) values ranged from 0.79 to 1.08, and 1.45 to 1.84, respectively. The AAESF values were further compared with those constrained using the signal at mass-to-charge 60 (m/z 60), a tracer for fresh biomass combustion, measured using aerosol chemical speciation monitor (ACSM) and aerosol mass spectrometry (AMS) instruments deployed at 16 sites. Overall, the AAESF values obtained from the two methods showed strong agreement, with a coefficient of determination (R2) of 0.78. However, uncertainties in both approaches may vary due to site-specific sources, and in certain environments, such as traffic-dominated sites, neither approach may be fully applicable.Peer reviewe
Towards a Framework for Intelligent Sampling: Comprehensive Review of Challenges, AI Techniques, and Tools
No full textPublisher Copyright: © 2025 IEEE.Intelligent data sampling is an innovative method that enhances conventional data sampling procedures by utilizing machine learning and artificial intelligence approaches. In this manuscript, we deep dive into the scientific and grey literature to find the main challenges faced by data sampling and elaborate on the various AI techniques utilized to mitigate them. We identify key issues such as class imbalance, overfitting, computational inefficiency, and bias, which often hinder traditional sampling methods. Furthermore, we explore AI-driven techniques that have been integrated into the sampling process to address these challenges effectively. As a result, we propose a novel framework for intelligent sampling that incorporates an AI-powered recommender system. This system dynamically selects the most appropriate sampling technique based on the specific characteristics of the data and the needs of the predictive model. By automating and optimizing the selection of sampling methods, our framework aims to enhance model performance, improve resource efficiency, and adapt to diverse real-world applications.Peer reviewe
Engineered Protein-Based Ionic Conductors for Sustainable Energy Storage Applications
No full textPublisher Copyright: © 2025 The Author(s). Advanced Materials published by Wiley-VCH GmbH.Protein-based biomaterials offer sustainable and biocompatible alternatives to traditional ionic conductors, essential for advancing green energy storage and bioelectronic applications. In this work, a robust, intrinsically self-assembling repeat protein scaffold to enhance ionic conductivity through the selective incorporation of glutamic acids is engineered. These mutations increase the number of available protonation sites and promote the formation of well-defined charge pathways. The self-assembly properties of the system enable the propagation of molecular-level modifications to the macroscopic scale, yielding self-standing protein films with significantly improved ionic conductivity. Specifically, engineered protein-based films exhibit an order of magnitude higher conductivity than their unmodified counterparts, with a further ten-fold enhancement through controlled addition of salt ions. Mechanistic analysis shows that the conductivity enhancement originates from the intertwined contributions of proton transport, hydration, and ion diffusion, all promoted by engineered charged residues. Finally, films of the best-performing variant are integrated, as both separator and electrolyte, into a supercapacitor device with competitive energy storage performance. These findings highlight the potential of rational protein design to create biocompatible, sustainable, and efficient ionic conductors with the stability and processability required to be successfully integrated into the next generation of energy storage and bioelectronic devices.Peer reviewe
Study on quantum thermalization from thermal initial states in a superconducting quantum computer
Get PDFPublisher Copyright: © The Author(s) 2025.Quantum thermalization in contemporary quantum devices, in particular quantum computers, has recently attracted significant interest. However, there are few experimental results due to the difficulty in preparing thermal states in quantum systems. In this paper, we propose a protocol to indirectly address this challenge using only pure states. While our protocol does not solve the issue of thermal state preparation, it enables the equivalent study of their dynamics. Moreover, we experimentally validate our protocol using IBM quantum devices, presenting results that demonstrate unusual relaxation in equidistant quenches. We also assess the formalism introduced for the Quantum Mpemba Effect (QME), which provides a framework for comparing the dynamics of different thermal states, we do no observe any unusual behaviour in this case, which is consistent with the theoretical predictions for the system. This demonstration underscores that our protocol can provide an alternative way of studying thermal states physics when their direct preparation may be too difficult.Peer reviewe
Large Language Models for Structured Task Decomposition in Reinforcement Learning Problems with Sparse Rewards
No full textPublisher Copyright: © 2025 by the authors.Reinforcement learning (RL) agents face significant challenges in sparse-reward environments, as insufficient exploration of the state space can result in inefficient training or incomplete policy learning. To address this challenge, this work proposes a teacher–student framework for RL that leverages the inherent knowledge of large language models (LLMs) to decompose complex tasks into manageable subgoals. The capabilities of LLMs to comprehend problem structure and objectives, based on textual descriptions, can be harnessed to generate subgoals, similar to the guidance a human supervisor would provide. For this purpose, we introduce the following three subgoal types: positional, representation-based, and language-based. Moreover, we propose an LLM surrogate model to reduce computational overhead and demonstrate that the supervisor can be decoupled once the policy has been learned, further lowering computational costs. Under this framework, we evaluate the performance of three open-source LLMs (namely, Llama, DeepSeek, and Qwen). Furthermore, we assess our teacher–student framework on the MiniGrid benchmark—a collection of procedurally generated environments that demand generalization to previously unseen tasks. Experimental results indicate that our teacher–student framework facilitates more efficient learning and encourages enhanced exploration in complex tasks, resulting in faster training convergence and outperforming recent teacher–student methods designed for sparse-reward environments.Peer reviewe
The Practical Implications of Re-Referencing in ERP Studies: The Case of N400 in the Picture–Word Verification Task
No full textPublisher Copyright: © 2025 by the authors.Background: The selection of an optimal referencing method in event-related potential (ERP) research has been a long-standing debate, as it can significantly influence results and lead to data misinterpretation. Such misinterpretation can produce flawed scientific conclusions, like the inaccurate localization of neural processes, and in practical applications, such as using ERPs as biomarkers in medicine, it may result in incorrect diagnoses or ineffective treatments. In line with the development and advancement of good scientific practice (GSP) in ERP research, this study sought to address several questions regarding the most suitable digital reference for investigating the N400 ERP component. Methods: The study was conducted on 17 neurotypical participants. Based on previous research, the references evaluated included the common average reference (AVE), mean earlobe reference (EARS), left mastoid reference (L), mean mastoids reference (MM), neutral infinity reference (REST), and vertex reference (VERT). Results: The results showed that all digital references, except for VERT, successfully elicited the centroparietal N400 effect in the picture–word verification task. The AVE referencing method showed the most optimal set of metrics in terms of effect size and localization, although it also produced the smallest difference waves. The most similar topographic dynamics in the N400 window were observed between the AVE and REST referencing methods. Conclusions: As the most optimal regions of interest (ROI) for the picture–word elicited N400 effect, nine electrode sites spanning from superior frontocentral to parietal regions were identified, showing consistent effects across all referencing methods except VERT.Peer reviewe
Enabling 3D electrical stimulation of adipose-derived decellularized extracellular matrix and reduced graphene oxide scaffolds in vitro using graphene electrodes
Get PDFPublisher Copyright: © 2025 The Royal Society of Chemistry.Notwithstanding the demonstrated benefits of electrical stimulation in enhancing tissue functionality, existing state-of-the-art electrostimulation systems often depend on invasive electrodes or planar designs. This work exploits the versatility of graphene to fabricate biocompatible electrodes for the three-dimensional in vitro electrical stimulation of neural stem cells. A conductive green graphene-based ink was formulated and screen-printed as the bottom and top electrodes in a bottom-less standard culture well plate. Upon exposure to macrophages, although some oxidative stress was observed, this graphene-based ink did not elicit an increase in the pro-inflammatory cytokine IL-6. An analysis of the electrode impedance as a function of time and frequency was performed to optimize the 3D electrical stimulation. The efficacy of these graphene electrodes for electrically stimulating cells across 3D environments was investigated in scaffolds composed of a decellularized extracellular matrix and reduced graphene oxide, which had previously shown the capability to facilitate neuronal differentiation in vitro and to create a pro-regenerative microenvironment in vivo. Neural stem cells were seeded on these scaffolds and electrically stimulated with a 10 Hz bidirectional current signal of 200 μA for 1 hour daily. At the target frequency of 10 Hz, deemed advantageous for neural regeneration, a scaffold impedance below 800 Ω was ensured. The low-frequency 3D stimulation proved to enhance cellular mechanisms essential for the development of neuronal networks, including neuronal differentiation, neuritogenesis and neurite growth.Peer reviewe
A roadmap to a low-cost anion exchange membrane unitized regenerative fuel cell
No full textPublisher Copyright: © 2025 The Royal Society of Chemistry.A unitized regenerative fuel cell (URFC) is a device that converts and stores electricity generated from renewable sources into hydrogen, which may subsequently be converted back into electricity as needed. The commercialization of traditional proton exchange membranes and alkaline URFCs is hampered by the high cost of platinum group electrocatalysts and the differential pressure required to connect a URFC to renewable energy sources. The anion exchange membrane-based unitized regenerative fuel cell (AEM-URFC) is a promising option for large-scale renewable energy storage and hydrogen generation. It does not require a costly platinum metal catalyst, and it is readily incorporated into renewable energy systems. This is new technology in its infancy, and thus it requires a potential roadmap for sustainable growth, prior to its commercialization. This review describes recent advances made in the creation of AEM-URFC modules and their performance. It also presents comparisons with conventional technology and a brief economic analysis. The purpose is to summarize recent developments and then to recognise gaps in the existing literature. It concludes with a discussion of some challenges and suggestions for potential actions relevant to the development of long-lasting AEM-URFCs.Peer reviewe
Membership Inference Attacks Fueled by Few-Shot Learning to Detect Privacy Leakage and Address Data Integrity
Get PDFPublisher Copyright: © 2025 by the authors.Deep learning models have an intrinsic privacy issue as they memorize parts of their training data, creating a privacy leakage. Membership inference attacks (MIAs) exploit this to obtain confidential information about the data used for training, aiming to steal information. They can be repurposed as a measurement of data integrity by inferring whether the data were used to train a machine learning model. While state-of-the-art attacks achieve significant privacy leakage, their requirements render them infeasible, hindering their use as practical tools to assess the magnitude of the privacy risk. Moreover, the most appropriate evaluation metric of MIA, the true positive rate at a low false positive rate, lacks interpretability. We claim that the incorporation of few-shot learning techniques into the MIA field and a suitable qualitative and quantitative privacy evaluation measure should resolve these issues. In this context, our proposal is twofold. We propose a few-shot learning-based MIA, termed the FeS-MIA model, which eases the evaluation of the privacy breach of a deep learning model by significantly reducing the number of resources required for this purpose. Furthermore, we propose an interpretable quantitative and qualitative measure of privacy, referred to as the Log-MIA measure. Jointly, these proposals provide new tools to assess privacy leakages and to ease the evaluation of the training data integrity of deep learning models, i.e., to analyze the privacy breach of a deep learning model. Experiments carried out with MIA over image classification and language modeling tasks, and a comparison to the state of the art, show that our proposals excel in identifying privacy leakages in a deep learning model with little extra information.Peer reviewe
Practical deployment and validation of an IoT based semantic interoperability approach for industrial interoperability in smart manufacturing
Get PDFPublisher Copyright: © 2025Industrial 5.0 emphasizes human central, resilient, and sustainable manufacturing. Achieving this vision requires seamless interoperability between various systems on the shop floor, including both automated systems and human workers. However, fragmentation of industrial protocols hampers efficient data exchange, delaying intelligent automation and digitalization. This paper presents SHOP4CF smooth data exchange and intelligent automation vision through the use case validation of the Web of Things Interoperability Layer (WoT-IL) as a solution to bridge the communication gaps between the heterogeneous industrial protocols. everaging FIWARE's context broker and OpenAPI specifications, WoT-IL enables standardized, semantic interoperability across sensors, PLCs, and higher-level systems. The component was deployed and evaluated in two real-world industrial pilots, demonstrating its ability to reduce manual configuration, enhance system flexibility, and support vendor-agnostic integration. Validation used a structured framework that evaluated functionality, integration, process improvement, human factors, and worker acceptance. Results confirm that WoT-IL significantly improves interoperability and automation readiness, positioning it as a key enabler for scalable, human-aware, and future-proof industrial ecosystems. This position WoT-IL as an enabler to improve automation, reduce system fragmentation, and support a more efficient, connected, and worker-friendly industrial environment.Peer reviewe