Dynamic source domain selection: An adaptive EEG transfer learning framework for mitigating negative transfer.

A multi-platform analytical strategy for Atractylodes lancea authentication: Fusion of stable isotope, elemental, chromatographic, and spectroscopic profiles.

MSHLoRA: A multi-stage multi-scale diverse adapter framework for efficient expert-level adaptation of large language models

ASG-FU: An adaptive and secure grouping framework for federated unlearning

• Digital Twin integrates IoT quantum sensors for adaptive greenhouse lighting. • Real-time DT interactions enables energy-efficient and responsive light control. • RL optimizes controller gains under multi-zone variable PPFD and light conditions. • Energy use and electricity costs reduced versus baseline on/off strategy. • GA and RL compared over 12 adaptive control strategies in real greenhouse data. Greenhouse lighting is vital for plant growth and contributes to nearly 30% of operational costs. However, managing lighting in response to dynamic sunlight conditions and varying photosynthetic photon flux density (PPFD) requirements across crop types remains a major challenge, which results in excessive energy use. This paper presents a digital twin (DT) adaptive control framework for greenhouse lighting, leveraging quantum sensors and reinforcement learning (RL) to enable energy-efficient, multi-zone operation. The proposed system dynamically adjusts multi light-emitting diode (LED) intensities in the extended Photosynthetically Active Radiation spectrum to satisfy uniform PPFD thresholds (single-crop scenario) or differentiated PPFD thresholds (multi-crop scenario). A set of 12 adaptive control strategies was evaluated, employing Genetic Algorithm (GA) and RL optimizers to configure proportional–integral-derivative (PID) controller parameters as well as their PI and P subsets. Real world validation demonstrates that the RL-based PI control with shared coefficients (RL-PI (Eq)) delivers the most robust performance across scenarios, achieving an average mean error of 1.248 μ mol s − 1 m − 2 with a standard deviation of 10.661 μ mol s − 1 m − 2 . Compared to a baseline on–off controller, the proposed strategy reduces electrical energy consumption by 23.6% and energy-related costs by 23.2%, while maintaining precise PPFD regulation. These findings highlight the potential of DT adaptive control systems to advance sustainable, cost-effective, and scalable multi crop greenhouse lighting management.

AutoWAFuzzer: An adaptive framework for web application firewall penetration testing with multi-agent system and RAG-enabled reinforcement learning

LLM-SDaT: A knowledge-informed LLM framework for syndrome differentiation in TCM.

Selective knowledge distillation-based domain adaptation framework towards edge computing fault Diagnosis for high-speed train bogie

Matched-field processing is highly sensitive to environmental mismatch, yet most robust formulations emphasize static uncertainties more than time-evolving environmental forcing. This study examines a representative low-frequency shallow-water scenario in which wind-driven mixed-layer deepening reshapes the upper-ocean sound-speed profile and perturbs modal horizontal wavenumbers, producing accumulated phase errors, ambiguity-surface distortion, and systematic range bias. To organize these effects beyond a single operating point, a conditional modal phase-spread analysis is introduced to show how wind-driven degradation depends jointly on wind state, propagation range, frequency, and source depth relative to the mixed layer. A physics-coupled particle filter (PC-PF) is then proposed, in which wind speed is treated as a dynamic hidden state and estimated jointly with source range through an embedded reduced-order environmental model. Broadband numerical experiments are used to assess mechanism and tracking performances. For the representative storm-evolution scenario considered here, a conventional static-model broadband Bartlett processor develops kilometer-scale range errors, whereas the proposed PC-PF substantially reduces the root mean square error and preserves track continuity. The formulation is intended as a reduced-order, acoustically informed framework for dynamic environmental adaptation in time-varying conditions.

Cytosolic PRDX1 acts as an extramitochondrial sink to set mitochondrial H2O2 levels and enable resilience to chronic mitochondrial oxidative stress.

Recruitment for clinical trials in acute care settings, including intensive care units (ICUs), emergency departments, and peri-operative environments, presents distinctive ethical and practical challenges. These include time-sensitive decision-making, impaired patient capacity, complex consent processes, and reliance on surrogate decision-makers (SDMs). Despite growing attention to research ethics, there is a limited synthesis of how these challenges impact patient consent and trial recruitment. Following PRISMA-ScR guidance, a systematic search of PubMed/MEDLINE (2010-2025) was conducted, supplemented by a targeted ClinicalTrials.gov search to identify ongoing and completed trials. Data were extracted and classified into overarching domains to map the breadth of barriers and proposed solutions. Fifteen studies were included, spanning randomised controlled trials, qualitative studies, surveys, and reviews. Key barriers were: (1) timing and capacity constraints in high-pressure environments; (2) comprehension challenges with technical consent documents, especially in minority populations; (3) regulatory variability limiting harmonisation of alternative models (e.g., deferred and staged consent); and (4) under-representation of vulnerable groups, leading to selection bias. Nine additional randomised clinical trials identified in ClinicalTrials.gov explored innovative consent tools such as digital aids and AI chatbots; two had published results. Promising strategies included multimodal consent, culturally tailored materials, early SDM engagement, and patient and public involvement (PPI). Acute-care and anaesthetic trials require adaptive consent frameworks that balance ethical rigour with operational feasibility. Adoption of inclusive recruitment strategies and harmonised international standards may enhance both equity and efficiency in future critical-care research.

Enhancing hospital drug and medical supply request processes through a clinically prioritized and demand-variability-aware adaptive decision support framework.

The high rate of electric vehicles (EVs) development has motivated the issues of peak load congestion, data privacy, scalability, and secure energy coordination in smart electric mobility networks. The traditional centralized EV charging management systems have weaknesses of privacy leakage, single point failure, lack of real time flexibility and lack of trust in the transaction. This paper proposes a Privacy-preserving Edge -Trust -Adaptive Learning Framework (PETAL-Grid), an AI-based federation blockchain model to support adaptive and privacy-preserving energy coordination. The key goal of this study is to attain scalable, real-time and secure EV charging coordination through the integration of federated artificial intelligence, edge-based demand intelligence and blockchain enabled trust management. The proposed framework allows joint demand learning without the need to exchange raw data, real-time adaptive charging based on edge intelligence, and transparent and tamper-proof energy transactions based on smart contracts. The PETAL-Grid workflow comprises of local data collection, edge-based demand forecasting, federated model aggregation, adaptive load coordination, and blockchain-based transaction validation. The results of the simulation show that PETAL-Grid can attain 18% peak load reduction, 17% efficiency of energy utilization, and 98-99% transaction security, which are better than the centralized and the baseline models. The results validate that PETAL-Grid is a scalable, reliable and dependable solution to sustainable smart electric mobility networks.

LADA: A label-aware framework for cross-domain sentiment classification.

An adaptive isogeometric framework for topology optimization based on a reaction-diffusion equation

Umbrella protocols have recently come to be widely used in clinical trial designs. However, the value of this approach in public health is less well recognized. The coronavirus disease 2019 (COVID-19) pandemic highlighted the need for rapid, reliable and scalable evaluation of diagnostic technologies. In the United Kingdom of Great Britain and Northern Ireland, this need prompted the development of an umbrella research protocol enabling multiple clinical evaluation studies of similar designs to be undertaken under a single overarching preapproved ethics and governance framework. We describe the development, implementation and evolution of this protocol, which was designed to support timely assessment of the performance of in vitro diagnostic devices and associated testing approaches in various settings. The umbrella protocol allowed studies to be started quickly during the pandemic, reduced administrative burdens, supported regulatory submissions and enabled prospective collection of samples for surveillance. While the system described reflects British governance structures, the principles underpinning this approach, including proportionality (ensuring oversight requirements are appropriate to the risk level), standardization and preapproved flexibility, are applicable to many settings. The protocol now forms part of the United Kingdom's wider pandemic preparedness structure and illustrates how preapproved, adaptable research frameworks can accelerate evidence generation during outbreaks. The world is now assessing lessons from the COVID-19 pandemic and it is timely to consider how research systems can support innovative designs such as umbrella protocols. We therefore summarize lessons learnt and practical considerations to support other countries seeking to adopt similar approaches within their own ethical and regulatory systems.

Physics integrated adaptive residual learning framework for traffic state estimation

Adaptive risk analysis framework for network-Level moving target defense under adversarial intelligence uncertainty

An enhanced model-free adaptive predictive control framework for parallel mechanism in hydraulic humanoid robots

Meta-Learning Enhanced Multi-Source Domain Adaptation for zero-calibration motor imagery EEG decoding.