A fuzzy deep learning approach for liver lesions detection and classification in big data context

Understanding the behavior of Atlantic salmon ( Salmo salar ) in commercial sea cages is critical for developing effective strategies to manage parasites. It is particularly relevant since many preventive measures, such as lice skirts, snorkel cages, and submerged systems, aim to limit encounters between salmon and lice in the upper part of the water column. Furthermore, behavior is also a recognized welfare indicator, with deviations from normal patterns potentially signaling stress or disease. Despite its importance, behavioral baselines remain poorly defined, especially in large-scale production environments. This study investigates the use of a mechanical 360-degree single-beam scanning sonar (Ping360) to monitor the vertical and horizontal distribution of post-smolt Atlantic salmon (510 g on average) in a commercial sea cage during the first six weeks after sea transfer, a period associated with elevated stress and mortality. Unlike traditional single beam echosounders, which offer limited horizontal coverage, the Ping360 provides a broader field of view. The fish monitored in this study started to feed the day after their transfer and initially occupied the upper five meters of the water column, with no marked differences in day-night distribution outside some agitation and spreading in the cage center when feed was provided. Some schooling patterns along the cage walls formed after five days and became the main daytime behavior after two weeks. Feeding behavior also became more structured over time, with fish aggregating below the surface in the center of the cage and dispersing gradually as they became satiated. Salmon did not display a particular avoidance of the bottom or center of the cage. Additionally, cage deformation due to elevated current velocities and waves were visible on the sonar output. No major changes in fish distribution were observed under these conditions, except for one instance where salmon seemed to swim 1–2 m further away from the surface. Our findings demonstrate the potential of this low-cost, user-friendly sonar to capture informative behavioral and cage deformation data in large-scale aquaculture settings. • This is the first use of a 360° sonar to monitor salmon distribution in large sea cages. • Both vertical and horizontal distributions can be monitored. • Main behavioral patterns like schooling, feeding, and use of space were identified. • Cage deformation under currents was visible in the sonar data. • This tool shows promise for behavior and cage deformation monitoring.

Revealing the divergences between LLM-simulated and human takeover decision-making in ADS-equipped HGV operations.

Abstract We present Ly α , He ii , and C iv observations of seven z ∼ 2 radio-loud quasars observed using the Keck Cosmic Web Imager and compare it to observed radio jet emission using archival Very Large Array and Atacama Large Millimeter/submillimeter Array radio observations. We detect 80–120 kpc diameter Ly α and 10–40 kpc He ii and C iv emission around the targets. We find the Ly α emission to be brighter within the inner 30 kpc by factors of 2–10 compared to other literature samples. We reproduce the trend for increased total luminosity for a larger on-sky area, but find our targets tend to be brighter for a given area when compared to literature observations, even when adjusting for the observational sensitivity. We infer that the He ii and C iv is likely powered by quasar photoionization, with the ionizing radiation likely escaping along the radio jet axis, which is aligned with the He ii and C iv emission. The observations agree with a two-component model of the circumgalactic medium (CGM), where the inner CGM (<30 kpc) is directly influenced by the host galaxies, whereas gas motions in the outer CGM (>30 kpc) are governed by turbulence and the larger-scale environment.

The rapid proliferation of Internet of Things (IoT) devices has significantly increased the attack surface for large-scale botnet operations. While previous research, including detailed analyses using Cowrie and IoTPOT frameworks, has studied IoT botnet behavior, these studies often rely on retrospective datasets, isolated protocol analyses, or hard-to-replicate setups. This paper addresses that gap with two main contributions: a structured review of ten influential IoT security studies from the USENIX Security Symposium and a confirmatory empirical experiment deploying Cowrie and IoTPOT honeypots simultaneously on a Microsoft Azure cloud-based virtual machine. Unlike earlier studies that focus on single protocols or large-scale environments, this work acts as a validation study, confirming well-known IoT botnet behaviors, including credential brute-force attacks, Mirai-style commands, and Telnet dominance, using real-time attack data collected from a reproducible, affordable cloud environment that simulates known IoT vulnerabilities (such as CVE-2016-10401, CVE-2017-17215, and CVE-2014-9222). Rather than revealing new attack methods, this study explicitly verifies the persistence of behaviors first documented almost ten years ago. The data indicates that attackers continue to exploit basic authentication flaws and reuse long-standing command sequences, confirming that core IoT vulnerabilities remain prevalent despite a decade of security research. It also highlights the ongoing gap between research progress and industry implementation. The analysis situates these findings within the broader evolution of IoT botnets, from early centralized command-and-control structures like Mirai to more resilient peer-to-peer networks that use anonymized channels and target high-wattage devices for power-grid manipulation. This study shows that small, cloud-based honeypots are valuable for continuous threat monitoring, model validation, and security assessments, providing a practical, reproducible approach for ongoing IoT security research.

ABSTRACT Effective wayfinding design is crucial for navigating complex public spaces, especially in large-scale healthcare environments, where it significantly impacts patients, families, and medical professionals. Due to the spatial complexity of modern hospitals, effective wayfinding requires a design-led approach that prioritises the diverse needs of end-users. This paper presents a case study involving the redesign of signage at the Shanghai Oncology Hospital (SOH). Facilitated by the Research Centre for Digital Innovation and Humanities Education (RCDIHE) at Macau University of Science and Technology, the study utilised focus groups and co-design workshops to explore navigation barriers. The process engaged medical professionals to generate iterative prototypes aimed at enhancing the health-seeking experience. Results demonstrate that applying a Participatory Design-led Engagement (PD-E) framework in conjunction with the CEO (Co-design, Exploration and Outcome) framework effectively informed and validated the strategic placement and design of guide boards, proving the value of co-design within the context of a Chinese public hospital.

The emergence of dense, mission-driven aerial networks supporting the low-altitude economy presents unique communication and security challenges, including extreme channel dynamics and severe cross-tier interference. Traditional reactive communication paradigms are ill-suited to these environments, as they fail to leverage the network’s inherent predictability. This paper introduces <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">predictive communication</i>, a novel paradigm transforming network management from reactive adaptation to proactive optimization. The approach is enabled by fusing predictable mission trajectories with stable, large-scale radio environment models (<italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">e.g.</i>, radio maps). Specifically, we present a hierarchical framework that decomposes the predictive cross-layer resource allocation problem into three layers: strategic (routing), tactical (timing), and operational (power). This structure aligns decision-making timescales with the accuracy levels and ranges of available predictive information. We demonstrate that this foresight-driven framework achieves an order-of-magnitude reduction in cross-tier interference and enables proactive security against threats such as jamming and spoofing, laying the groundwork for efficient, resilient, and secure low-altitude communication systems.

(Visual) Simultaneous Localization and Mapping (SLAM) remains a fundamental challenge in enabling autonomous systems to navigate and understand large-scale environments. Traditional SLAM approaches struggle to balance efficiency and accuracy, particularly in large-scale settings where extensive computational resources are required for scene reconstruction and Bundle Adjustment (BA). However, this scene reconstruction, in the form of sparse pointclouds of visual landmarks, is often only used within the SLAM system because navigation and planning methods require different map representations. In this work, we therefore investigate a more scalable Visual SLAM (VSLAM) approach without reconstruction, mainly based on approaches for two-view loop closures. By restricting the map to a sparse keyframed pose graph without dense geometry representations, our ‘2GO’ system achieves efficient optimization with competitive absolute trajectory accuracy. In particular, we find that recent advancements in image matching and monocular depth priors enable very accurate trajectory optimization without BA. We conduct extensive experiments on diverse datasets, including large-scale scenarios, and provide a detailed analysis of the trade-offs between runtime, accuracy, and map size. Our results demonstrate that this streamlined approach supports real-time performance, scales well in map size and trajectory duration, and effectively broadens the capabilities of VSLAM for long-duration deployments to large environments.

In the digital age, verification of academic credentials has emerged as a serious problem with the growing challenge of document forgery and dependence on centralized system. Conventional methods are slow, susceptible to mistakes, and at risk of being attacked. In this paper, we present a scalable and secure blockchain-based certificate authentication system that provides a tamper-proof and transparent validation for both certificate issuers and the issued certificates. The system uses Ethereum smart contracts and SHA-256 hashing to guarantee immutability and integrity of the data. A consensus-driven voting procedure is utilized to authenticate the trusted institutions, which prevents having a single point of trust. In addition, to improve performance, a Bloom Filter-based search optimization method is employed, resulting in a dramatic reduction on the time of certificate verification, especially for large scale environment. The proposed framework is designed with strong security and privacy guarantees to withstand attacks, e.g., Sybil attacks and 51% attacks, while maintaining user's data privacy by only depositing cryptographic hashes on the blockchain. The simulation results illustrate lower transaction costs and higher verification efficiency than other existing solutions. The technique offers a secure, decentralized, and cost-effective method to authenticate certificates in both academic and professional fields

We investigate the relation between supermassive black holes (SMBHs), their host galaxies, and large-scale dark-matter halos by combining broad-line X-ray active galactic nuclei (AGNs) from the XMM--XXL and Stripe,82X surveys with galaxies from VIPERS and SDSS/Stripe,82. Building on the homogeneous host-galaxy catalogue developed in Paper I, we test whether AGNs with a given black-hole mass, M_̊m BH, occupy the same or different large-scale environments as non-AGN galaxies with statistically indistinguishable host properties. We characterised the empirical M_̊m BH--M_⋆ distribution of the AGN sample. A shallow scaling between M_̊m BH and stellar mass, M_⋆, is present, but with a large intrinsic scatter influenced by flux-limited selection and virial-mass uncertainties. The ratio ̊m BH /M_⋆ declines with increasing M_⋆ over the sampled range. Overmassive and undermassive AGN subsets, defined relative to this empirical trend, exhibit distinct median host properties consistent with partially non-synchronous SMBH and M_⋆ growth. We then selected AGNs in two M_̊m BH intervals, 8.0 łe łog(M_ ̊m BH /M_⊙) &lt; 8.5 and 8.5 łe łog(M_ ̊m BH /M_⊙) &lt; 9.0, and constructed galaxy control samples matched in M_⋆, SFR, and sSFR ($= $) using a multivariate nearest-neighbour procedure. Using AGN--galaxy cross-correlation functions, we inferred characteristic halo masses for AGNs and matched galaxies in each bin. The AGNs with 8.0 łe łog(M_ SFR M_⋆ ̊m BH /M_⊙) &lt; 8.5 occupy halos statistically indistinguishable from those of their controls, indicating no detectable environmental dependence at these masses once host properties are controlled. In the higher-mass bin, 8.5 łe łog(M_ ̊m BH /M_⊙) &lt; 9.0, we find a mild indication that AGNs may reside in more massive halos than the matched non-AGN galaxies. The inferred difference is sim0.4,dex but remains formally consistent within the uncertainties. If confirmed with larger samples, this may indicate that halo-scale processes become increasingly relevant only at the highest M_̊m BH, while at lower masses AGN environments remain indistinguishable from those of inactive galaxies with similar host properties.

Network Intrusion Detection Systems (NIDSs) are critical to ensuring the resilience of modern digital infrastructures. Although traditionally deployed in large-scale corporate environments, the expanding threat landscape requires the integration of robust security measures into Small Office/Home Office (SOHO) and Edge-of-Things (EoT) networks. However, these environments often face significant constraints in terms of specialized hardware and technical expertise. This article presents R-Snort, an open-source NIDS based on Snort 3, optimized for low-cost Raspberry Pi 5 hardware. Its multi-agent architecture enables distributed deployment with centralized traffic analysis and cross-agent attack correlation, while an intuitive web interface simplifies alert visualization and system management for non-expert administrators. Its main contributions are: (1) a performance-optimized NIDS agent achieving 1 Gbps throughput; (2) a distributed multi-agent architecture enabling centralized event correlation and detection of multi-vector attacks; and (3) an IaC-based automated deployment framework with an intuitive web interface, democratizing professional-grade security for SOHO and EoT environments.

Vehicle environments generate diverse acoustic signals from engines, braking systems, road interactions, and surrounding traffic. Analysing these sounds provides valuable insights into vehicle behaviour and operational status. With the rapid growth of intelligent transportation systems and datadriven automotive technologies, automated vehicle audio analysis has emerged as a key research area. Traditional monitoring systems relied on manual inspection or basic signal processing techniques, which were limited in handling large-scale audio data and complex acoustic environments. These approaches also struggle with the high volume of data generated by modern sensor-equipped vehicles, highlighting the need for intelligent, data-driven solutions. This research presents a machine learningbased vehicle audio event classification framework that integrates advanced feature extraction and classification techniques. The system processes raw audio signals and extracts deep acoustic representations using the Waveform Language Model (WavLM), a transformer-based feature extraction model. These extracted feature vectors are used to train multiple classifiers, including CatBoost Classifier (CBC), Histogram Gradient Boosting Classifier (HGBC), Extra Tree Classifier (ETC), and the proposed Tree-Based Generalized Additive Model (TGAM). The models are evaluated using standard performance metrics such as accuracy, precision, recall, and F1-score. Experimental results show that the TGAM model significantly outperforms the other classifiers. It achieves an accuracy of 99.92% for main class classification and 99.58% for sub class classification, demonstrating its effectiveness in recognising complex vehicle audio events. This framework enhances intelligent vehicle monitoring systems by enabling accurate and efficient acoustic signal analysis

Intelligent Traffic Engineering in Software-Defined Networks for Enhanced QoS in Large-Scale IoT Environments

Mobile Unmanned Aerial Vehicles (UAVs) forming Flying Ad Hoc Networks (FANETs) offer promising applications, but dynamic network structures, limited resources, and potential single points of failure create security challenges. While cluster-based data aggregation, where data is collected and combined at Cluster Heads (CHs) before transmission, improves efficiency, traditional techniques can compromise data privacy. This paper introduces SSDBFAN, a scalable and secure cluster-based data aggregation framework for Flying Ad Hoc Networks (FANETs). The proposed approach integrates the Frilled Lizard Optimization Algorithm (FLOA) for efficient cluster head selection with blockchain technology and post-quantum cryptographic techniques, including lattice-based homomorphic encryption and the Chinese Remainder Theorem, to ensure privacy-preserving data aggregation. Additionally, a hybrid online/offline signature mechanism is employed to achieve secure and efficient authentication with reduced computational overhead. The performance of the proposed framework is evaluated using NS-3 simulations under varying network sizes. Experimental results demonstrate that SSDBFAN significantly improves communication efficiency, reduces computational cost, and enhances network stability compared to existing schemes. Furthermore, scalability analysis with up to 500 UAV nodes confirms that the proposed framework effectively controls blockchain overhead, including bandwidth consumption, consensus latency, and storage requirements. Comparative evaluation with existing optimization algorithms shows that FLOA achieves superior performance in terms of cluster stability, delay, and throughput. These results validate the effectiveness of SSDBFAN as a scalable and security-aware solution for large-scale FANET environments.

Abstract—Cloud computing allows access to computing re-sources in a scalable and on-demand fashion, yet effective, resource allocation is also a major problem because of the dynamic workloads and unpredictable user demands. The com-mon allocation approaches, including the provisioning that is not dynamic and scheduling that is based on the rules, tend to lead to low utilization of resources, high latency, and high operational costs. To solve these problems, this paper suggests an AI-based solution to optimize the resources distribution in the cloud environment. The suggested system will use the methods of Artificial Intelligence, such as, but not limited to, Machine Learning, Deep Learning, and Reinforcement Learning, to predict the workload trends and dynamically distribute the resources. The predictive models are trained using historical data like CPU usage, memory consumption, and network bandwidth. These models help to pre-dict resource needs with accuracy and proactively and efficiently assign them. Reinforcement Learning also boosts decision-making as it continuously improves allocation policies through system feedback. Experimental findings show that the suggested solution can be used to better utilize resources and minimize latency and the overall system performance than conventional solutions. Scalability and adaptability in a large-scale cloud environment is also supported by the model. This paper has shown that the application of AI methods in cloud resource management is an efficient way to get effective, cost-efficient, and smart resource allocation. Keywords: Cloud Computing, Dynamic Resource Allocation, AI-based Scheduling, Reinforcement Learning, LSTM, Predictive Analytics, Virtualization, Quality of Service (QoS). Index Terms—component, formatting, style, styling, insert

Abstract Tropical cyclone (TC) genesis depends on the favorability of the large-scale environment. This study demonstrates how zonal asymmetry shapes tropical atmospheric and oceanic circulations and alters environmental favorability for TC genesis implied by the genesis potential index (GPI). Two idealized coupled climate models are examined, one with (Ridge) and one without (Aqua) an oceanic meridional boundary—Aqua (Ridge) has a zonally symmetric (asymmetric) climate. Ridge has a zonally asymmetric GPI distribution that peaks near the equator over the warm pool with high sea surface temperature (SST) and moves poleward near the cold tongue with low SST, resembling observed GPI pattern in the South Indian and Western South Pacific basins. Despite warmer SSTs near the warm pool, Ridge produces a less favorable environment for TC genesis than Aqua because of enhanced vertical wind shear associated with the Walker circulation and insufficient absolute vorticity due to the equatorward shift of ITCZ, in addition to overall unfavorable conditions near the cold tongue. In both configurations, the seasonal cycle of GPI follows the seasonal shift of low vertical wind shear. Composite anomalies of GPI during different phases of ENSO-like variability on Ridge are calculated separately to illustrate the change of environmental favorability for TC genesis induced by the interannual variability. GPI decreases (increases) and shifts equatorward (poleward) near the cold tongue during El Niño (La Niña) associated with a weaker (stronger) Walker circulation. Our results demonstrate the value of an idealized model hierarchy for exploring global circulation and TC interactions (including ENSO impacts) within climate models.

Abstract Feedback from radio AGN jets is regularly implemented into contemporary models of galaxy evolution to offset radiative cooling in the large-scale environments in which they typically reside. While previous studies suggest that the total kinetic power output from radio AGN is sufficient for this purpose, many have relied on jet-power estimation from radio luminosities using generalised scaling relations that neglect additional information such as source size and environment. We here infer the cosmic evolution of radio AGN kinetic jet powers using a physically motivated semi-analytic model for the first time. Initial analysis on a sample of 619 radio AGN at z &amp;lt; 2.5 from LoTSS Deep Field and International LOFAR Telescope images of the Lockman Hole implies a population dominated by short-lived sources typically of lower jet power. After incorporating weighting towards shorter lifetimes in the inference models, we utilise ELAIS-N1 and Boötes LoTSS Deep Field data to expand our analysis to a much larger sample of 5,187 objects, deriving jet kinetic luminosity functions and integrated kinetic luminosity densities for the radio AGN population out to z = 2.5. In broad agreement with previous results in the literature, we find the total power output per comoving volume to be ∼1032–1033 W Mpc−3 across the full redshift range, with some suggestions of moderate positive evolution from z = 0–1 and little evolution from z = 1–2. These values are compatible with expectations from some cosmological models, providing strong evidence for the viability of feedback from radio AGN jets across cosmic time.

Abstract Despite numerous studies on the drivers of southern African precipitation, important gaps remain in understanding the role of the southern continental thermal low. This paper investigates the Kalahari Thermal Low (KTL), its connections to the Angola Low (AL) and Botswana High (BH), and its influence on regional precipitation. Although the AL and the BH form at different pressure levels during austral summer, they are dynamically linked: a deepened AL is typically associated with a weakened BH. These two atmospheric features shape the large-scale environment in which the KTL develops as a semi-permanent, thermally–driven system. The KTL originates over Angola in September, intensifies as it migrates poleward, peaks over the Kalahari in January and recedes rapidly to disappear near its origin by March. Our results show that KTL intensity is strongly and negatively correlated with precipitation across southern Africa. This relationship is explained by the leading mode of variability, which is primarily controlled by the development of a heat dome over southern Africa, rather than by anomalous mid-tropospheric advection of dry air into the convective region, as previously suggested. During warm KTL phases, anomalous surface heating induces tropospheric subsidence that evolves into a heat dome, creating conditions conducive to drought. Conversely, cold KTL phases are associated with a reduced mid-tropospheric divergence, which favours anomalous moisture convergence that deepens further the convection and increases rainfall.

With the continuous growth in semiconductor demand, semiconductor manufacturers have expanded the capacity of fabrication (FAB) facilities. Consequently, an Automated Material Handling System (AMHS), which controls Overhead Hoist Transport (OHT) vehicles, has become larger and more complex. In large-scale FAB environments characterised by dense track networks, frequent congestion often leads to bottlenecks, decreasing overall productivity. However, traditional static routing approaches for each OHT fail to effectively manage real-time congestion, resulting in traffic being concentrated on specific areas and causing operational inefficiencies. To address this issue, we propose a reinforcement learning-based Dynamic Link Weight Control (DLWC) method. By partitioning the AMHS layout into multiple areas, the DLWC controls the link weights that influence OHT route selection, leading OHTs to choose less congested paths. Simulation experiments demonstrate that the proposed DLWC method outperforms conventional rule-based approaches in terms of both throughput and lead time, validating the practical effectiveness of congestion control in large-scale AMHS environments.

Our study investigates how individuals cooperate in strategic dilemmas under varying levels of conflict, with a focus on the role of empathy and reciprocity. Using a modified version of the 2×2 Conflict Game, we examine whether dispositional empathy (measured through the Interpersonal Reactivity Index) predicts cooperative behavior, and whether sequential decisions elicit reciprocal responses. Our results show that higher levels of conflict reduce cooperation, while cognitive empathy and observed cooperation from others increase prosocial choices. Notably, this experiment was conducted with a large, diverse sample of university students in a nonincentivized setting, using minimal infrastructure during a compulsory citizenship education course. Although the study was not designed to evaluate pedagogical outcomes, our implementation demonstrates that behavioral experiments can be successfully adapted to large-scale classroom environments. This contributes methodologically to the growing literature on scalable experimental designs in education (crowd experiments), while advancing theoretical understanding of empathy and reciprocity in conflict-based decision-making.