134125 research outputs found
Sort by
Privacy-Preserving Automated Deep Learning for Secure Inference Service
No full textAutomated deep learning (AutoDL) aims to automatically discover optimal architectures of deep neural networks (DNNs) for secure inference without the studies for time-consuming and error-prone manual design. Privacy concerns have increasingly motivated the studies for privacy-preserving AutoDL (PrivAutoDL), where DNN architectures are searched directly on encrypted data without revealing the client’s confidential inputs and well-trained DNN architectures. However, existing studies encounter problems in achieving a balance between provable security and efficiency while avoiding significant degradation of model utility. To tackle these problems, we design a privacy-preserving AutoDL scheme, named 2PCAutoDL, utilizing a two-party (two non-colluding cloud servers) computation model. Based on the two-server model, efficient and secure computation protocols are customized layer by layer to protect DNN models associated with client’s data. In particular, we reduce the computational overhead of secure DNN: our optimized protocols achieve 1.34× ∼ 2.05× speedup for linear layers and 1.33× ∼ 45× speedup for nonlinear layers, compared to a range of existing secure implementations in the literature. Moreover, our fresh alternative to approximate Softmax avoids the drawbacks of approximating exponential operation and yields slightly higher accuracy under appropriate configurations. The security of 2PCAutoDL is formally analyzed under the semi-honest adversary model. Extensive experiments demonstrate that the searched models from 2PCAutoDL improve the inference accuracy by 0.6% on MNIST and by 0.5% on CIFAR10 when compared to state-of-the-art (SOTA) PrivAutoDL.No Full Tex
The appearance of microblade technology at the onset of Marine Isotope Stage 2 in Northeast China
No full textLate Pleistocene lithic miniaturization, particularly the emergence, development and spread of microblade technology in East and Northeast Asia, has received substantial attention. Here we present a comprehensive study of the Xishantou site in Northeast China, dated to 28.3–27.4 cal ka BP. The lithic assemblage is characterized by intensive exploitation of local tuff and rhyolite and the coexistence of diversified technical systems, including core-flake and laminar reduction strategies (i.e. blade, bladelet and microblade removals). Microblades and bladelets of the site were primarily detached from burin-cores by soft-stone-hammer percussion and formed an independent reduction sequence distinct from blade reduction. Therefore, the Xishantou site provides unequivocal evidence for the appearance of microblade technology in Northeast China at the onset of Marine Isotope Stage 2 (MIS 2). This abrupt technological shift within the long-standing core-flake tradition of Northeast China may be closely linked to the eastward migration and dispersal of northern Eurasian populations, coupled with climate and environmental deterioration during the transition from MIS 3 to MIS 2. Subsequently, southward migration of populations equipped with microblade technologies to more hospitable regions was recorded in several sites during the Last Glacial Maximum (LGM), for example, at the site of Helong Dadong in the Changbaishan Mountains. These findings provide critical new insights into the emergence and development of microblade technology in Northeast China. They also shed light on the population dynamics and their adaptive strategies in response to fluctuating climate and environment during the Late Pleistocene.No Full Tex
Patient perspectives on research gaps in cluster headache
No full textObjective: This study was undertaken to identify gaps in cluster headache management, highlight patient-prioritized research needs, and assess patient interest in, and preferences for, clinical trial participation. Background: Many people with cluster headache still lack effective treatment options to control or prevent attacks. There is a critical need for more studies, particularly clinical trials, in this field. To design and conduct successful trials, it is essential to identify priority research areas, allocate resources effectively, and ensure patient engagement and support. Methods: This study was an online survey conducted among Australian adults with self-reported cluster headache. Participants were recruited using a multi-channel approach, including direct outreach by clinicians, support from patient advocacy groups, and broad social media distribution. It collected data on demographics, treatment experiences, and perspectives on future research, including research priorities, and preferred outcomes and interventions. Additionally, participants' interest in joining clinical trials was assessed to help identify potential candidates for future studies. Results: Of the 219 individuals who began the survey, 17 (8%) were excluded due to providing no responses beyond demographic data or reporting no cluster headache diagnosis by a healthcare professional. The final sample consisted of 202 participants, with an average age of 46 years, 77% aged 25–54 years, 55% male, 72% had been living with cluster headache for more than 10 years, and 29% reported attacks occurring almost every month throughout the year. A quarter of participants had not followed up with a healthcare provider for cluster headache management when they completed the survey. Among those who sought care (n = 145 [72%]), general practitioners were the most frequently consulted (86%), followed by neurologists (66%). Treatments were considered “not at all effective” or “somewhat ineffective” by 35% of all participants, while 27% reported only partial effectiveness. The main treatment challenges were ineffectiveness (74%), side effects (54%), cost (53%), and difficulties with access (39%). Among the 202 participants, 126 (62%) indicated interest in participating in future cluster headache trials, while 26 (13%) responded with “maybe.” Psilocybin was the highest-ranked treatment in terms of participants who were “very interested,” with 66% selecting this option. The combined proportion of participants who were “very interested” or “interested” was 84% for combination therapies, 82% for psilocybin, 71% for medical devices, and 66% for anti-CGRP treatments. Conclusion: Participants with reported cluster headache highlighted inadequate treatment options, emphasized the need for further research, and expressed interest in future clinical trials, particularly those involving psilocybin or combination therapies.No Full Tex
The opportunities and challenges of online learning in middle level education: Perspectives of students, parents, and school staff in the post-pandemic era
No full textThe COVID pandemic was a major disruptor to education systems globally and locally. These disruptions resulted in emergency transitions to remote online learning, which differed from traditional face-to-face learning. As middle level education has entered the post-pandemic era, it is important to explore the perspectives of key collaborators (e.g. students, teachers, and parents) to identify current opportunities and challenges of online learning. This will provide important insights for middle grades educators and inform future strategies for ensuring positive online learning experiences, as post-pandemic recovery is important, especially in the middle years. Therefore, the present study used semi-structured interviews to gather the collective perspectives of three collaborators [Parents (n = 4); students (n = 6; grade 6, aged 11–12 years) and school staff (e.g. teachers) (n = 5)] from one suburban school in Southeast Queensland, Australia. The data allowed triangulation of middle grades perspectives on post-pandemic opportunities and challenges of online learning. Through a reflexive process of thematic analysis, eight key themes emerged: (a) Opportunities: ways of learning, self-learning, online capabilities, and online support; and (b) Challenges: technical issues, online distractions, screen use, and online safety. These findings showed that multiple opportunities exist that offer positive online learning experiences in the middle grades. However, parents, students, and school staff still grapple with several current challenges of online learning at school, and practical recommendations are provided to address these challenges.No Full Tex
MalModel: hiding malicious payload in mobile deep learning models with black-box backdoor attack
No full textMobile malware has become one of the most critical security threats in the era of ubiquitous mobile computing. Despite the intensive efforts from security experts to counteract it, recent years have still witnessed a rapid growth of identified malware samples. This could be partly attributed to the newly-emerged technologies that may constantly open up under-studied attack surfaces for adversaries. One typical example is the recently-developed mobile machine learning (ML) framework that enables storing and running deep learning (DL) models on mobile devices. Despite obvious advantages, this new feature also inadvertently introduces potential vulnerabilities (e.g., on-device models may be modified for malicious purposes). In this work, we propose a method to generate or transform mobile malware by hiding malicious payloads inside DL models’ parameters based on a strategy considering four factors (layer type, layer number, layer coverage, and the number of bytes to replace). Utilizing the proposed method, we can run malware in DL mobile applications covertly with little impact on the model performance (i.e., as little as 0.35% drop in accuracy and at most 39ms latency overhead). We can successfully trigger malicious functions, such as getting SMS records and screenshots in a real-world application. The generated malware can evade state-of-the-art detection techniques (i.e., none detected by VirusTotal), and the malware-based attack exhibits high practical feasibility (i.e., successfully attack 41% of the apps with on-device DL models). Our work should alert security experts on malware injection attacks on mobile devices, and further raise more awareness towards the deep-learning-assisted attacks in the mobile ecosystem.No Full Tex
Surveying the scene: A review of how landscapes are valued
No full textPlanning theory and practice has placed considerable attention on the conservation of ecologically and culturally significant landscapes (ECSLs). A broad range of valuation systems have been proposed and implemented. This study performed a systematic literature review of 112 articles addressing the valuation of ECSLs to identify the landscape features that are most valued and why. The findings indicate that planning theorists and practitioners tend to follow material valuation approaches that prioritise the conservation and management of tangible landscape values. This trend has been traditionally driven by research and practice based in Europe, North America, Australia and Japan, and more recently from China. It follows more structured valuation systems such as the ecosystems services approach proposed by many international frameworks. In contrast, local landscape users assign value to landscape features depending on context-based experiences and aspirations that are underpinned by non-material valuation systems. The consideration of such intangible values could determine the extent to which stakeholders engage in landscape conservation. There is a need to develop more comprehensive valuation systems that can accommodate both material and non-material landscape values.No Full Tex
Watching Myanmar: A bibliography of works on Myanmar (Burma) by Andrew Selth, 1983–2025
Full text linkThis bibliography was produced by the Griffith Asia Institute (GAI) at Griffith University, Brisbane. It provides a comprehensive listing of works published between 1983 and 2025 by Andrew Selth that examine Myanmar (formerly Burma) and its place in the regional and global order.
The publication brings together more than four decades of scholarship, including books, peer-reviewed monographs, research papers, working papers, journal articles, commentaries and selected online contributions. It supersedes earlier listings and offers an updated, structured record of Selth’s contributions to Myanmar studies.
This bibliography is intended as a reference resource for researchers, students, policymakers and practitioners seeking to better understand the political, strategic and historical dynamics of Myanmar.Full Tex
The Impact of Power Distance Belief on Consumers' Preference for Perceived Core and Peripheral Destination
No full textThe role of power distance belief (PDB) as a cognitive lens shaping the mental processes underlying product selection has received limited attention. Drawing on construal level theory (CLT), this research examines how consumers' PDB influences their process- and outcome-focused product evaluation, which in turn shapes their product preferences. Using a mixed-method approach across four studies, we empirically test this model in a tourism context, focusing on consumers' choices between core and peripheral destinations. Specifically, low-PDB consumers tend to adopt a process-focused evaluation and therefore prefer perceived peripheral destinations, whereas high-PDB consumers are more likely to adopt an outcome-focused evaluation and prefer perceived core destinations. We also identify two boundary conditions—process priming and consumers' photo-sharing intentions. Together, these findings illuminate how PDB shapes the cognitive pathway underlying product selection, offering important implications for understanding individual-level cultural influences on consumer behaviour.No Full Tex
Mesoporous Materials for Electrochemical Biosensors: From Broad Structure to Silica Film
No full textMesoporous materials, characterized by tunable pore structures, large surface areas, and versatile surface properties, have become essential for advancing electrochemical biosensors. Their unique structural features enable efficient immobilization of biomolecules, enhancing electron transfer and amplifying signal transduction, which are vital for precise and selective detection of analytes within complex biological environments. This review provides a thorough discussion of scaffold-based mesoporous materials, including mesoporous silica, carbon, metals, metal oxides, metal-organic and covalent organic frameworks. We are focusing on their design strategies, synthesis techniques, and functional integration into biosensing systems. We emphasize recent innovations that leverage these materials for health monitoring and clinical diagnostics, targeting clinically relevant biomarkers such as glucose, nucleic acids, proteins, and metabolites linked to specific diseases. Moreover, we explore the role of mesoporous silica thin films in the operation of miniaturized biosensors, highlighting their potential in wearable and point-of-care applications. By systematically evaluating performance metrics and addressing reproducibility and scalability challenges, this review outlines crucial pathways for future research. Additionally, we discuss how advances in materials science, surface enhancement, and device integration are driving the development of next-generation real-time, non-invasive, and personalized diagnostic tools.Full Tex
Health Students' Experiences in an Innovative University Teaching Model: The Good, the Bad, the Challenging
No full textWhile universities worldwide transitioned online for necessity, many now continue blended or online programs due to growing interest. Higher education is also exploring innovative intensive learning and teaching models. One example is the Southern Cross Model (SCM), promoting focused learning using immersive teaching, active learning, and guided learning to encourage autonomy and promote community. An anonymous online survey explored international and domestic health students’ experience of the SCM, identifying current satisfaction with and self-confidence in learning and subjective perception of the SCM. A total 105 students completed the survey, 69.5% domestic (n=73) and 30.5% international (n=32). Domestic students reported significantly greater learning satisfaction (Mdn = 25.00) than international students (Mdn = 20.50; U = 701.50, z = -3.46, p <.001), as well as higher self-confidence than international students (Domestic Mdn = 36.00; International Mdn = 32.00; U = 862.00, z = -2.15, p = .032). All students acknowledged the intensity and challenge of the SCM, but international students found it more overwhelming and isolating. In contrast, domestic students reported enhanced motivation and engagement. These findings emphasise student experiences in intensive teaching pedagogies are evaluated, providing a foundation for tailored support programs supporting all students to thrive in new educational approaches.Full Tex