134125 research outputs found
Sort by
Handling data sparsity and model poisoning attacks in federated sequential recommender systems
Get PDFFederated sequential recommendation (FedSeqRec) allows many user devices to train a shared recommender without sending raw interaction histories to a central server, which is important for privacy. However, existing FedSeqRec methods still suffer from two key limitations: (1) most users have very short or sparse histories, especially when they only show interest in a few items for a short period, leaving the model with too little data to understand their preferences; and (2) in a sequential setting it is normal for interests to change suddenly, but the model may misinterpret these abrupt changes as anomalies. In this paper, we propose FORTRESS, a Federated cOntrastive RobusT REcommender for Sequential Systems, designed to address these limitations. To tackle the first issue, FORTRESS generates augmented versions of local interaction sequences on each client, so that the model can observe more plausible behaviour patterns and learn user preferences more reliably even when histories are short or sparse. This extra flexibility also gives adversaries more room to manipulate the training signal, so we complement it with a popularity-aware server-side regularizer that discourages rare or suspicious items from drifting into the same embedding clusters as genuinely popular items. To tackle the second issue, we introduce a temporal regularization term that discourages abrupt changes in user representations across adjacent subsequences, allowing the model to adapt to short-term interest shifts while still preserving stable long-term tastes. Experiments on three real-world datasets show that FORTRESS improves recommendation accuracy for sparse and cold-start users and substantially reduces the success of strong model poisoning attacks compared with competitive centralized and federated baselines.Full Tex
Working with deep leverage points: Insights from an ARC project on sustainability in tourist accommodation
No full textNo Full Tex
Assessment of legacy and emerging contaminants in sediment of an Australian deepwater ocean outfall
No full textThis study evaluated the impacts of a wastewater treatment works (WWTW) in a regional city in Australia that releases both secondary-treated wastewater and resultant biosolids into an ocean outfall (depth: 25 m). Sediments (n = 64) were collected in both nort-heast (NE) and south-west (SW) directions at 10–20, 50–100, 200–500, and 1000–2000 m distance from the diffusers. The sediments were analysed for physicochemical properties, trace metals, microplastics (MPs), estrogenicity, and estrogenic endocrine disrupting chemicals (e-EDCs). The sediments were low in clay and organic content, but were enriched in fine sands. Most of the trace metal concentrations were below the detection limit of ICP-OES. Only the concentrations of Al, Fe, Cr, and Se were reportable, but were far below sediment quality guideline values. Average MP concentrations (size: >25–5000 μm) were also low (NE: 42 particles kg−1; SW: 54 particles kg−1) and were composed of 68% fibres (PET and PP) and 32% fragments (PP and PE). The estrogenicity (NE: 0.89 ng EEQ g−1; SW: 1.19 ng EEQ g−1) of the sediments was low to medium. Estrogenic compounds most responsible for the estrogenicity were 17β-estradiol (E2) and bisphenol A (BPA). All parameters were significantly higher near the diffusers (i.e. , 10–20 m) and e-EDCs, particularly E2 might pose high ecological risks to aquatic habitats at this distance. Overall, there was limited contamination in the ocean outfall sediment, which is most likely attributable to the effect of significant oceanic dilution, deficiency of clay and organic content, and high content of fine sand in sediment.Full Tex
Can a Peripheral Blood Marker for Airway Neutrophilia Be Identified in Children with Bronchiectasis?
No full textBackground: Airway bacterial infection and inflammation are often present in children with bronchiectasis. Systemic inflammation has also been reported. Currently, there are no data on the association between systemic inflammatory markers with airway pathogens or neutrophilia in children with bronchiectasis. We aimed to define the bronchoalveolar lavage (BAL) pathogens (bacteria and viruses), and cytology in children with bronchiectasis and to explore any association between peripheral inflammatory markers and airway neutrophilia. Methods: Participants numbering 402, aged 15%) which was associated with the presence of H. influenzae (OR 2.03 95% CI 1.31–3.15, p = 0.002), S. pneumonia 2.41 (95% CI 1.36–4.29, p = 0.003), and Adenovirus (OR 2.06 95% CI 1.06–4.04, p = 0.033). Airway neutrophilia was associated with raised CRP (OR 2.26 95% CI 1.14–4.49, p = 0.019), but there were no other systemic inflammatory markers including monocyte/lymphocyte ratio, neutrophil/lymphocyte ratio, platelet/lymphocyte ratio, and platelet/mean platelet volume ratio. Conclusions: In children, there is an association between airway neutrophilia and raised CRP in bronchiectasis, but not with other peripheral inflammatory markers. There is a need to identify non-invasive inflammatory markers in children with bronchiectasis.Full Tex
Hums in the humus: opportunities and challenges for soil ecoacoustics
No full textSoil ecoacoustics is an emerging field and suite of tools that use sound and vibration to detect belowground biological activity. It offers a minimally invasive way to assess soil communities and ecosystem processes. Across biomes, we found that soil ecoacoustics is being used to detect organisms, quantify animal behaviour, monitor soil health, and assess restoration interventions. In this review, we show that ecoacoustic metrics reflect changes in soil fauna activity, disturbance impacts, and recovery trajectories. However, major challenges remain, including inconsistent terminology, limited understanding of sound propagation across soil types, difficulty separating biotic from abiotic signals, and a lack of standardised methods. Thus, we propose foundational standard operating procedures (SOPs) and identify how soil ecoacoustics could be integrated into global biodiversity monitoring frameworks.No Full Tex
Authentic reflection in clinical placement: Educator views on fostering reflective practitioners
Get PDFIntroduction: Reflection offers many benefits for medical students, improving self-awareness and integration of theory into real situations. However, like any skill, reflection must be learned and practiced. Reflective learning is also influenced by the workplace, particularly in students’ clinical training years. This study explored the factors perceived by educators to influence reflective learning amongst medical students undertaking a rural longitudinal integrated clerkship (LIC).
Methods: All educators within the LIC program were invited to participate. Educators come from both a medical and nursing background. Two focus groups (FG) were conducted. FG were facilitated by experienced researchers. Transcripts were analysed by two researchers using Braun and Clark’s approach to thematic analysis.
Results: Eleven educators participated in two focus groups (six medical educators; five nursing educators). Educators perceived that reflective learning was influenced at three levels: student, educator and institution. These findings aligned with workplace learning theory. Educator and institutional influences were identified as “affordances” which support or hinder reflection. Supportive affordances included labelling of reflection, multidisciplinary collaborative learning, and the longitudinal nature of the rural model. Hindering affordances included using artificial reflective assessments. Educators also identified characteristics of the individual student which may contribute to student decision-making about reflection.
Conclusion: This study provides a deeper understanding about educator experiences of reflective learning within LIC settings. The findings provide insights into the ways in which educators, educational institutions and student workplace settings may support or hinder reflective practice. This study also highlighted the potential for the LIC model to support reflective practice.Full Tex
Chemical inhibition of a bacterial immune system
Get PDFThe rise of antibiotic resistance motivates a revived interest in phage therapy. However, bacteria possess dozens of anti-phage immune systems that confer resistance to therapeutic phages. Chemical inhibitors of these anti-phage immune systems could be employed as adjuvants to overcome resistance in phage-based therapies. Here, we report a class of chemical inhibitors that selectively inhibit type II Thoeris anti-phage immune systems from diverse bacteria—including antibiotic-resistant pathogens, thereby sensitizing phage-resistant bacteria to phages. These inhibitors block the biosynthesis of a histidine-ADPR intracellular “alarm” signal by ThsB, thereby preventing ThsA from arresting phage replication. Chemical inhibition of the Thoeris defense improves the efficacy of a model phage therapy against a phage-resistant clinical isolate of P. aeruginosa in a mouse infection, suggesting a therapeutic potential. These findings demonstrate that the selective inhibition of anti-phage defense systems can improve the efficacy of therapeutic phages, suggesting a strategy to circumvent phage-therapy resistance.Full Tex
A Living Scoping Review of Universal Interventions for Promoting Relational Health in Childhood, Adolescence and Young Adulthood
No full textInvesting in relational health across childhood, adolescence and young adulthood not only promotes health and development within a generation, but may have cascading benefits to the next generation. Here we review the literature on universal interventions designed to promote relational health from childhood to young adulthood (4-24 years), just prior to the normative transition to parenthood and raising next generation offspring. This review was conducted in accordance with the JBI methodology for scoping reviews. Electronic databases (MEDLINE [EBSCOhost], PsycINFO [EBSCOhost], and Embase [EBSCOhost] databases) were searched using terms that combined concepts: (1) outcomes pertaining to child, family and community relational ecology; (2) childhood, adolescence, and young adulthood; (3) RCT study design; (4) universal prevention approach. This yielded 3,396 articles, of which 113 were eligible for inclusion. A further 12 articles were identified via expert knowledge resulting in 125 articles reporting on 85 universal interventions (including nine population interventions). Most (90%) interventions were designed for children and adolescents, and most (97%) targeted family, school and community microsystems including aggressive/disruptive behaviour, parenting, peer relationships, and social competence using mostly classroom/school and parenting/family interventions. The few mesosystem interventions focused mostly on family-school connections. Only nine made changes to exosystems, for instance through community coalitions. Key features of population trials included: (1) multiple components; (2) involvement of the community, and; (3) integration into existing service systems. Efforts to promote relational health could be strengthened by a focus on improving the the interlocking social infrastructure which enables relational health to flourish at the microsystem level.No Full Tex
Ask and focus more: Question-prompt uncertainty allocation for dual-controllable video captioning
No full textVideo captioning aims to generate natural language descriptions from video content via hierarchical architectures that capture key visual elements. Although entities, predicates, and syntactic structures are critical for coherent descriptions, existing controllable methods often struggle to dynamically emphasize salient information due to reliance on predefined entity lists and static fusion strategies. Moreover, uncertainty in aligning multimodal features (e.g., visual actions and textual semantics) further weakens model robustness. To address these challenges, we propose QPDC, a Dual-Controllable video captioning framework that introduces a Question-Prompt mechanism to adaptively guide caption focus. Unlike traditional entity-driven approaches, QPDC adopts a unified question-prompt strategy to maintain semantic consistency while enabling flexible attention shifts. Specifically, the Question-Prompt (Q-Prompt) module dynamically steers the model's focus according to query intent, emphasizing salient content without predefined constraints. The Selective Fusion Module (SFM) hierarchically integrates Q-Prompt guidance with local action features and global sentence context to support multi-granularity control. In addition, a Dirichlet-based strategy is employed to model uncertainty in multimodal weight allocation, improving adaptability under varying data conditions. Extensive experiments on MSVD and MSR-VTT show that QPDC achieves competitive performance among controllable video captioning methods and produces more diverse and user-preferred descriptions. Code will be available at https://github.com/XiaotaiCHEN/QPDC.git.No Full Tex
Elevated CO2 has a greater impact on surface soil carbon but warming on subsoil carbon: evidence from SOC accumulation and stability
No full textAims
Climate change, driven by elevated atmospheric CO₂ and warming, significantly influences soil organic carbon (SOC) dynamics, yet its combined effects on SOC accumulation, stability, and underlying mechanisms—particularly across soil profiles—remain poorly understood. This study aims to quantify the individual and interactive effects of elevated CO₂ and warming on SOC accumulation, stability, and their controlling factors across the 0–100 cm soil profile in a subtropical paddy, based free-air CO₂ enrichment (FACE) experiment for 12 years.
Methods
We conducted a decadal-scale FACE experiment in a rice paddy. Four treatments were applied: ambient conditions (CK), elevated CO₂ (600 ± 20 ppm, CE), canopy warming (+ 2 °C, W), and their combination (CW). Soil samples were collected from 0–100 cm depth to analyze SOC content, fractions (labile and recalcitrant), and stability indices. Statistical and modeling approaches, including random forest analysis, were used to identify controlling factors of SOC dynamics.
Results
Net SOC storage in the 0–100 cm soil profile by 20.3 Mg ha−1 under elevated CO2 alone but decreased by 13.0 Mg ha−1 with warming alone. The combined effect of elevated CO2 and warming was a loss on SOC stocks of 6.23 Mg ha−1. Elevated CO2 increased the recalcitrant organic carbon fraction (ROC, especially above 20 cm) and decreased the labile organic carbon fraction (LOC, especially below 20 cm). Warming alone slightly decreased ROC but significantly decreased LOC. The carbon pool management index (CPMI) declined under all climate change treatments, indicating a reduction in SOC pool quality, with warming exerting a stronger negative effect than elevated CO₂. Random forest analysis highlighted the pivotal roles of vegetation, soil and microbial processes in SOC dynamics. Soil chemical properties were more important than soil physical properties, and microbal necromass C made a more important contribution to SOC than plant quality and quantity.
Conclusions
This site-level study suggests that under future climate scenarios, the tested paddy soil may act as a net C sink under elevated CO₂ but as a net C source under warming. These contrasting responses highlight the need for further research across diverse sites to accurately predict regional C dynamics in subtropical paddy systems.No Full Tex