134125 research outputs found
Sort by
Cooperative dual-atomic-site catalysis enables highly efficient oxygen reduction reaction and long-life Zn-air battery
No full textRational design and synthesis of electrocatalysts that efficiently catalyze oxygen reduction reaction (ORR) is critical to the development of fuel cells and Zn-air battery (ZAB). However, it remains the most significant obstacle due to the potential-dependent energy barrier for oxygen intermediates (OOH*, O* and OH*) adsorption/desorption on single atom active sites. Here we report a novel deprotonated 2-aminoterephthalic acid (H2BDC-NH2) coordinated compound precursor-mediated method for synthesizing dual-atomic-site electrocatalyst (FeNi-DSAs/NSs) that facilitate ORR kinetics. The experimental and theoretical calculation results verify that the Fe and Ni cooperative effect induces a distinct negative shift of the d-band center for Fe, thereby reducing the adsorption strength of oxygen intermediates on Fe sites. Additionally, the introduced Ni species decrease the energy barrier of the rate-determining step (RDS), thus accelerating catalytic ORR process. Consequently, FeNi-DSAs/NSs exhibits excellent ORR performance in 0.1 M KOH solution with a very large half-wave potential (E1/2) of 0.92 V and a remarkable lifespan, maintaining stability for over 90 h. In aqueous Zn-air battery (A-ZAB), this electrocatalyst enables a high-power density (254 mW cm−2), and can maintain stable charge/discharge for more than 1600 h with negligible potential gap fluctuation. Notably, for quasi-solid-state Zn-air battery (QSS-ZAB) with FeNi-DSAs/NSs, it shows excellent cycle life for over 110 h and high energy conversion efficiency at 0.5, 1.0, and 2.0 mA cm−2, respectively. This work not only provides a novel design strategy for excellent catalysts, but also highlights the potential application of non-precious metal electrocatalysts in renewable energy conversion and storage apparatuses.No Full Tex
Generative AI in academic research activities: The hidden side of self-detrimental consumption
No full textGenerative AI (GenAI) is increasingly embedded in academic research activities undertaken by researchers (including research-active educators) and research students. While GenAI can raise efficiency, it may also foster self-detrimental consumption for short-term convenience that erodes long-term research integrity and capability. To map this “hidden side”, we conducted a netnography of discussions on X-platform (formerly Twitter) by self-identified researchers, research-active educators and research students (between October and November 2024; Study 1), alongside semi-structured interviews with 19 Australia-based researchers (aged 19–45; Study 2). Across the data, we identified five key themes: user misuse, environmental facilitators, usage barriers, GenAI limitations, and challenges, along with related sub-themes. Integrating both studies, we propose the GenAI Self-Detrimental Consumption (GAI-SDC) framework, which explicates how these factors interrelate within academic research contexts. The framework offers a focused lens for analyzing GenAI-related behaviors by examining how factors interact in academic research activities. The practical contribution includes actionable strategies from the framework, providing tangible measures for institutions, researchers, and developers to mitigate self-detrimental use and promote responsible GenAI integration in academic research activities.No Full Tex
Holding Hope Over Time: Evaluating the Enduring Impact of the AIMS Suicide Prevention Model
No full textThis two-year follow-up study evaluated the sustained impact of the AIMS suicide prevention pathway, a structured, short-term intervention delivered by the Wellbeing Team (WBT) in a large Australian metropolitan mental health service. The retrospective cohort included 232 individuals who engaged with the WBT for at least 2 weeks and were discharged to primary care. Re-presentation to public mental health services was examined across three thresholds: any service contact (Condition 1), secondary care involvement (Condition 2) and sustained case management (Condition 3). At 24 months, 66.4% of participants had not re-presented to any public mental health service. Only 8.2% required further engagement with secondary services, and just 3.0% transitioned into longer-term case management. Survival analysis showed no statistically significant differences based on prior service contact, suggesting the AIMS pathway was broadly effective across diverse clinical histories. These findings indicate that AIMS supports not only crisis stabilisation but sustainable recovery. Its dual emphasis on clarity and growth is foundational: clarity through structured clinical actions, Assessment, Intervention, Monitoring and Step Up/Down (AIMS) and growth through guided discovery, values alignment, and peer support. The model is further strengthened by integration of the SAFE framework for dynamic safety planning and the HOPE framework for values-based recovery, both of which are embedded in the Safe Life Guide mobile application. By meeting individuals where they are and addressing both immediate safety and long-term purpose, AIMS offers a system-level reimagining of crisis care, one that reduces dependency, restores direction, and fosters life in motion.No Full Tex
Maximizing and Satisficing in Job Search: A Person-Centred Approach
No full textThe existing literature calls for greater clarity on how different job search strategies influence outcomes and wellbeing. In Study 1 (N = 1052; Mage = 25.48; 41.9% females), we developed and validated the Maximizing and Satisficing Job Search Scale (MSJSS). Results demonstrated its associations with existing measures of general maximizing and satisficing tendencies and job search outcomes (e.g., choice regret, number of offers, and outcome satisfaction). Using the MSJSS, based on another sample that was split randomly into a test group (N = 1010; Mage = 25.64; 34.5% females) and a validation group (N = 1010; Mage = 25.63; 35.0% females), Study 2 revealed a 4-profile solution: maximizers, satisficers, neutrals, and adaptive satisficers. Individuals in the maximizers group obtained significantly more job offers than the other groups, whereas the satisficers received the least. Job seekers in the neutrals were the least satisfied and most regretful of the four profiles. The adaptive satisficers reported more job offers and similar levels of outcome satisfaction when compared to the satisficers. Implications for theory, job search practice, and future research are discussed.No Full Tex
Complementary Joint Learning for Weakly Supervised Solar Panel Mapping in Aerial Images
Get PDFClassical alternative training schemes for weakly supervised learning rely on an initialization-refinement process, which is prone to performance degradation due to the inherent noise and inductive bias in pseudo labels (PLs). To address this challenge, we propose a novel Complementary Joint Learning Framework (CJLF), which leverages the complementary strengths of diverse PL types and substantially reduces the adverse effects of inherent noise and inductive bias in PLs. CJLF employs a shared encoder and a two-stream decoder architecture, with each stream serving for a distinct purpose. The Object Segmentation Stream (OSS) is tasked with producing integral predictions with well-defined boundaries, leveraging a cross-scale attention accumulation strategy to resolve object ambiguities in complex scenes. The Object Localization Stream (OLS) aims to identify target objects accurately. To foster the collaboration between these two streams, we propose an Attention Fusion Module (AFM) that integrates self-attention features from OSS and cross-stream attention features. Additionally, a Subspace Contrast Constraint is introduced to improve discrimination in ambiguous regions, ensuring OSS robustness against PL noise and strengthening the complementary interaction between the streams. The effectiveness of CJLF is validated on two challenging aerial datasets for solar panel mapping, AerialGDA2020 and DualGMS. Experimental results, including comparisons with state-of-the-art methods and extensive ablation studies, demonstrate the superiority of CJLF.Full Tex
CLIP-Powered Domain Generalization and Domain Adaptation: A Comprehensive Survey
Get PDFAs machine learning evolves, domain generalization (DG) and domain adaptation (DA) have become crucial for improving model robustness across diverse environments. Contrastive Language-Image Pretraining (CLIP) plays a central role in these tasks, offering strong zero-shot capabilities that allow models to operate effectively in unseen domains. Yet, despite CLIP's growing influence, no comprehensive survey has systematically examined its applications in DG and DA, underscoring the need for this review. This survey provides a unified and in-depth overview of CLIP-driven DG and DA. Before reviewing methods, we establish precise and complete scenario definitions covering source accessibility (SA vs. SF), source number (SS vs. MS), and label relations (CS, PS, OS, OPS), forming a coherent taxonomy that structures all subsequent analyses. For DG, we categorize methods into prompt optimization techniques that enhance task alignment and architectures that leverage CLIP as a backbone for transferable feature extraction. For DA, we examine both source-available approaches that rely on labeled source data and source-free approaches operating primarily on target-domain samples, emphasizing the knowledge transfer mechanisms that enable adaptation across heterogeneous settings. We further provide consolidated trend analyses for both DG and DA, revealing overarching patterns, methodological principles, and scenario-dependent behaviors. We then discuss key challenges such as realistic deployment scenarios, LLM knowledge integration, multimodal fusion, interpretability, and catastrophic forgetting, and outline future directions for developing scalable and trustworthy CLIP-based DG and DA systems. By synthesizing existing studies and highlighting critical gaps, this survey offers actionable insights for researchers and practitioners, motivating new strategies for leveraging CLIP to advance domain robustness in real-world scenarios. A continuously updated list of related works is maintained at:https://github.com/jindongli-Ai/Survey-on-CLIP-Powered-Domain-Generalization-and-Adaptation.Full Tex
How user-generated food photos influence tourists' restaurant visits: The effect of aesthetics, familiarity, and regulatory focus
No full textRelying on others’ social media posts to select dining venues has become increasingly common. Drawing on the heuristic-systematic model and regulatory focus theory, this research explores how food aesthetic appeal and food familiarity of user-generated photos (UGPs) in foodstagramming influence tourists’ dining decisions. The roles of mimicking desire and the individual's promotion- or prevention-focused motivational orientation are also investigated using an experimental design. The results indicate that tourists exhibit a greater desire to mimic dining choices when UGPs feature food with high aesthetic appeal, particularly when the food is unfamiliar. Additionally, regulatory focus moderates this effect: individuals with a promotion focus exhibit stronger mimicking desire and higher restaurant visit intention in response to appealing UGPs. This research contributes to the literature on foodstagramming by identifying key influential factors and target tourist segments. It also provides insights for marketers aiming to enhance social media strategies through cost-effective foodstagramming content.No Full Tex
Mapping Tree Cover Patterns in an Urban Arboretum from Multispectral Drone Imagery Using Pixel-Based Classification and Object-based Image Analysis
No full textRemote sensing methods are valuable in mapping small urban forests. Their small size and the species heterogeneity in these forests, however make it difficult to apply traditional aerial and space borne remote sensing data and analysis methods. High resolution imagery from drones can be used for the characterization of small urban forests. But finding efficient and accurate methods of analysing the large volume of data from such imagery remains a challenge. In this study, we compare pixel-based supervised random forest classification and object-based classification of tree cover patterns in a small arboretum forest in south-east Queensland, Australia. Tree cover in this study refers to the total area of forest land covered by trees. Low tree cover areas are defined as locations that are sparsely covered with trees while high tree cover areas are those that are covered by tree canopies above 30%. A multispectral drone imagery with five bands (Blue, Green, Red-Edge, Red, and Near Infra-red -NIR) from September 2023 was used in this study. Additionally, vegetation indices, including Normalized Difference Vegetation Index (NDVI), Normalized Difference Red-Edge Index (NDRE), Green Normalized Difference Vegetation Index (GNDVI), Normalized Difference Red Green index (NDGR) index and Enhanced Vegetation Index (EVI) were computed and used in the image analysis. Specifically, three approaches were adopted in our analyses. Firstly, random forest classification of drone imagery at pixel level and secondly, object-based image segmentation and classification of the imagery. Finally, deep learning method was used to detect tree crowns in the study area. Results show a strong similarity between the two methods of analysis and imply that properly trained machine learning models running at pixel-level could produce accurate results that are comparable to OBIA-generated results of small urban forest characterization. Further, the deep learning method detected 891 contiguous tree crowns in the arboretum forest. The outcomes from our robust applied comparison of these methods, including the deep learning, demonstrate that fine-resolution imagery can be used to characterize small urban forest both at an individual tree crown level and also to show the patterns of tree cover at the forest level.Full Tex
Adaptive multi-stage fusion of hyperspectral and LiDAR data via selective state space models
No full textMulti-source Remote Sensing (RS) image classification has gained significant attention due to its ability to overcome the limitations of single-source data, leading to improved classification performance. However, the joint classification of hyperspectral images (HSI) and Light Detection and Ranging (LiDAR) data poses significant challenges due to their pronounced spectral– spatial heterogeneity, necessitating advanced fusion techniques to exploit their complementary attributes effectively. This paper introduces a novel hierarchical multi-stage fusion framework that integrates HSI and LiDAR features using a selective state space model, known for its linear complexity and ability to model long-range dependencies. The proposed model comprises three key stages. The early fusion stage employs multi-scale feature extraction with exponentially sized pooling kernels (20 , 2 1 , 2 2 ) to reconcile spatial resolution disparities. The middle fusion stage features a Local Visual State Space (LocalVSS) block with window-based directional scanning, paired with a CrossModal Attention Fusion (CMAF) module using adaptive gating to dynamically capture intricate inter-modal relationships. The late fusion stage leverages a ConcatMamba block for robust, sequence-driven integration. A dynamic weighted fusion mechanism with learnable weights synthesizes these outputs, optimizing task adaptability. Comprehensive experiments on widely used HSI-LiDAR datasets reveal that our proposed model achieves superior classification accuracy and efficiency, outperforming state-of-the-art CNN-, Transformer-, and Mamba-based models. The source code will be publicly available at https://github.com/zhangyiyan001/ AMSFN.No Full Tex
Visualizing the Reaction Products of Sodium Rhodizonate with Lead(II) or Barium(II) Salts: New Insights into Forensic Gunshot Residue Detection
No full textEstablishing the presence of gunshot residue (GSR) is a crucial investigative insight used by forensic scientists to place a discharged firearm in the vicinity of shooters or victims in the conduct of crime scene reconstructions. The most prevalent presumptive test for identifying GSR on objects or locations is the sodium rhodizonate test, which produces a visible red or orange coloration upon reaction with lead or barium, respectively. Despite widespread use since the 1950s, the identity of the chromophores produced when sodium rhodizonate reacts with GSR metals remains poorly characterized. This work provides new insights and solid-state structures of metal rhodizonate coordination species. Additionally, a host of new and unexpected chromophore candidates were identified, providing a more complete picture of the chemical complexity underpinning GSR testing. A multi-instrumental approach was undertaken to support the characterization of five lead, four barium, and one organic dye structures with single crystal diffraction studies, thereby linking these solid-state structures to solution-state behavior. This work assigned coloration of key solution- and solid-state reaction species using UV–visible and absorbance spectroscopy, and enabled the interpretation of solution-state reaction behavior from in situ mass spectrometry analysis of the mixtures, thereby providing unprecedented insights into the chemistry that underpins the sodium rhodizonate test for GSR.Full Tex