116320 research outputs found
Sort by
Quantum-Assisted Mangrove Forest Mapping Algorithm
No full textMangrove ecosystems have a high potential for carbon sequestration, making accurate mangrove mapping (MM) increasingly important for environmental monitoring. However, current benchmark index-based methods primarily focus on water/vegetation-related spectral bands while overlooking other potentially informative bands. To optimize the use of all available information, we utilize channel attention mechanisms in our framework to extract features from every channel, including those often overlooked yet informative ones. This approach enhances the accuracy of mangrove pixel classification. Accordingly, we further incorporate a quantum feature extraction (QFE) module to capture novel features. Unlike traditional convolutional neural networks (CNNs), which rely on affine mappings, quantum neural networks (QNNs) employ unitary operations to extract quantum features. Therefore, QFE captures radically new quantum features distilled by the QNN from a fundamentally different perspective, which provides additional information for decision-making. Specifically, we fuse two different channel attention features derived from both the CNN and QNN to generate the final classification results. Our experimental results indicate that integrating the CNN and QNN channel attention modules does yield promising outcomes, thereby demonstrating the potential of introducing the QNN to advance state-of-the-art MM techniques
Blast loading prediction from methane-air explosion in long straight tunnels
Get PDFLong, straight underground tunnels are vital to modern infrastructure, including resource extraction, transportation, and hydropower, but their confined nature heightens the risks posed by methane-air explosions. This study presents a dimensionless predictive model for evaluating blast loading resulting from methane-air explosions in long, straight tunnels, integrating numerical simulations, parametric analyses, and an artificial neural network (ANN)-based approach. By adopting dimensionless input and output parameters, the model becomes inherently scalable to various tunnel sizes and configurations, eliminating the influence of scale effects and ensuring a broad range of applicability. A validated CFD model using FLACS provided the database for the ANN training, capturing the transition from deflagration to shock wave formation as well as key pressure duration and magnitude characteristics. The ANN-based model, accounting for cross-sectional area and shape, tunnel length, fuel length, blockage ratio and obstacle spacing, demonstrated acceptable predictive capability with R2 values above 0.95 and most predictions within a ± 30 % error margin. Validation against large-scale experiments confirmed its reliability and practicality. This model significantly reduces computational costs and time, offering an efficient tool for predicting methane-air explosion loads in underground tunnels
Cachexia, Anorexia and Feeding Difficulties in Palliative Care Patients
No full textSymptoms related to anorexia (loss of appetite), cachexia (muscle loss and wasting), and dysphagia (reduced ability to eat) are devastating for a person with a palliative illness and their families. They are common, especially as illness advances, and occur in both malignant and nonmalignant illness. Not only do the conditions serve as a visual reminder of chronic disease, they also affect the psychology governing the social aspects of eating. For many years, anorexia, cachexia, and dysphagia as distinct conditions were overlooked by healthcare professionals and researchers as the focus was on the primary chronic illness instead (Fearon et al., Nat Rev Clin Oncol 10(2):90–99, 2013). Clinical research, particularly in the field of cancer cachexia, has gained new insight into the pathogenesis and management of the conditions, including anorexia and dysphagia. This chapter will define the conditions and outline the epidemiology, etiology, and pathophysiology and assessment and management of these symptoms
Numerical Simulation of Deep Mixed Columns Beneath the Heavy-Haul Railway Embankment
Get PDFOne approach used to modify loose sand settlement under railway embankment involves deep mixed columns (DMCs). The geometrical properties of DMCs, such as the length, diameter, and spacing between adjacent columns, affect the distribution of load, failure mechanism, and ultimate resistance of a foundation constructed above soil reinforcement by this method. Numerical analysis is used as an alternative to laboratory modelling to minimize the cost and simplify the creation of complex DMC geometries in the lab. In this research, 3D Plaxis models are used to examine the influence of DMCs constructed under the embankment of a heavy-haul railway. The model used in this study has been calibrated and adjusted based on historical data from experimental tests on silty sand with low relative density. The study results offer details on how the bearing capacity of loose soil below train embankments is affected by the length and spacing of columns under static and dynamic loads
DPC: Dual-Prompt Collaboration for Tuning Vision-Language Models
Get PDFThe Base New Trade off BNT problem universally exists during the optimization of CLIP based prompt tuning where continuous fine tuning on base target classes leads to a simultaneous decrease of generalization ability on new unseen classes Existing approaches attempt to regulate the prompt tuning process to balance BNT by appending constraints However imposed on the same target prompt these constraints fail to fully avert the mutual exclusivity between the optimization directions for base and new As a novel solution to this challenge we propose the plug andplay Dual Prompt Collaboration DPC framework the first that decoupling the optimization processes of base and new tasks at the prompt level Specifically we clone a learnable parallel prompt based on the backbone prompt and introduce a variable Weighting Decoupling framework to independently control the optimization directions of dual prompts specific to base or new tasks thus avoiding the conflict in generalization Meanwhile we propose a Dynamic Hard Negative Optimizer utilizing dual prompts to construct a more challenging optimization task on base classes for enhancement For interpretability we prove the feature channel invariance of the prompt vector during the optimization process providing theoretical support for the Weighting Decoupling of DPC Extensive experiments on multiple backbones demonstrate that DPC can significantly improve base performance without introducing any external knowledge beyond the base classes while maintaining generalization to new classes Code is available at https github com JREion DP
Clinical Trials and Regulatory Issues of Nanocarriers Employed in the Treatment of Chronic Obstructive Pulmonary Disease
No full textIn the 21st century, chronic obstructive pulmonary disease (COPD) is one of the major global health concerns characterized by high prevalence and substantial morbidity and mortality due to long-term exposure to irritants like smoke, pollution, genetic factors and respiratory infections. Chronic respiratory failure is a leading outcome, and it is linked with a significant economic burden resulting from healthcare costs and lost productivity. COPD is a significant healthcare challenge due to limited treatment options. Nanocarriers, like nanoparticles and liposomes, offer promise for targeted drug delivery to the lungs, potentially enhancing treatment effectiveness and safety. In regular clinical practice, nanocarriers require careful consideration and adherence to regulations. This includes evaluating safety, potential toxicity, and impact on patient outcomes. However, the achievement of real-world application in regular clinical practice necessitated the development of harmonized regulatory environments. Advancing COPD care globally requires thorough clinical trials, navigating complex regulations, and proving effectiveness in diverse patient populations. Researchers, investors, and regulatory agencies are actively collaborating to tackle challenges in developing safe and effective nanocarriers for COPD treatment. This chapter addresses the present state of clinical trials and the regulatory landscape, along with future prospects for nanocarriers employed in COPD treatment. This chapter also highlights emerging market and regulatory trends as well as proposes solutions to enhance the efficiency of the regulatory process in the context of COPD
Supplementary material to "Land Surface Model Underperformance Tied to Specific Meteorological Conditions"
No full textERS guideline recommendation on airflow for breathlessness: the pitfalls of applying GRADE evidence ratings to complex non-pharmacological interventions.
No full textWe welcome the new European Respiratory Society clinical practice guideline on symptom management for adults with serious respiratory illness [1], which we hope will increase attention to breathlessness and other under-recognised and under-treated symptoms. Net benefit from airflow for breathlessness is undersold by the conclusion “very low overall certainty of evidence” in the recent ERS clinical practice guideline on symptom management for adults with serious respiratory illness https://bit.ly/3ZtMGc
Improved assessment of post-fire recovery trajectory of forests in Amazon's protected areas
Get PDFProtected areas (PAs) in Amazon forests are vital in preserving tropical forest ecosystems and mitigating forest degradation. However, the increasing frequency and severity of fires in these regions necessitate a comprehensive understanding of post-fire vegetation recovery trajectories, which is essential to evaluate the effectiveness and resilience of PAs in the face of ongoing climate change. Recovery trajectories under natural conditions remain uncertain, as unregulated human settlements often interfere with or influence the recovery process, skewing the actual recovery rates detected by satellite remote sensing. To tackle this issue, we examined 2990 MODIS-derived fire events in eastern Amazon PAs from 2001 to 2020. We assessed the effectiveness of multi-source Earth observation data and the eXtreme Gradient Boost machine learning model to distinguish burned areas undergoing natural recovery (natural recovery areas) from areas that are permanently converted to other uses (permanently converted areas). We then analyzed greenness recovery rates and canopy structure recovery trajectories across all burned areas, natural recovery areas, and permanently converted areas. Greenness recovery rates were derived from Landsat data, while canopy structure recovery was assessed using GEDI lidar-derived metrics and the space-for-time substitution approach. Our model achieved an overall classification accuracy of 87.90 %, accurately differentiating natural recovery areas (n = 1944) from permanently converted areas (n = 1046). The differing patterns of post-fire greenness recovery rates and structure recovery trajectories highlight the importance of this distinction. In natural recovery areas, significant recovery of structural traits such as relative heights (RHs), canopy cover (CC), and plant area index (PAIs), was observed, returning to their pre-disturbance levels over a 20-year period. Notably, metrics related to understory recovery and plant vertical space use, such as PAI values across the entire vertical strata, exhibited stronger recovery rates than height-related metrics like RHs, highlighting their utility in characterizing complex ecosystem recovery processes. These findings demonstrate the potential and necessity of using multi-source Earth observation data to distinguish different post-fire vegetation recovery processes. This distinction improves our understanding of ecological recovery rates and the successional dynamics of post-fire forests under natural conditions, offering new opportunities to explore their biogeographical distribution, recovery rate variabilities, and impacts on carbon sequestration and ecosystem resilience
Data sources:
Access Repository Dashboard
Do you manage OPUS? Access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard!