Search Results (1,395)

Explainability and interpretability have emerged as essential considerations in machine learning, particularly as models become more complex and integral to a wide range of applications. In response to increasing concerns over opaque “black-box” solutions, the literature has seen a shift toward two distinct yet often conflated paradigms: explainable AI (XAI), which refers to post hoc techniques that provide external explanations for model predictions, and interpretable AI, which emphasizes models whose internal mechanisms are understandable by design. Meanwhile, the phenomenon of concept and data drift—where models lose relevance due to evolving conditions—demands renewed attention. High-impact events, such as financial crises or natural disasters, have highlighted the need for robust interpretable or explainable models capable of adapting to changing circumstances. Against this backdrop, our systematic review aims to consolidate current research on explainability and interpretability with a focus on concept and data drift. We gather a comprehensive range of proposed models, available datasets, and other technical aspects. By synthesizing these diverse resources into a clear taxonomy, we intend to provide researchers and practitioners with actionable insights and guidance for model selection, implementation, and ongoing evaluation. Ultimately, this work aspires to serve as a practical roadmap for future studies, fostering further advancements in transparent, adaptable machine learning systems that can meet the evolving needs of real-world applications. Full article

This study investigates the spatiotemporal evolution of oil released during the Agia Zoni II shipwreck in the Saronic Gulf in 2017, employing the OpenOil module of the OpenDrift framework. The simulation integrates oceanographic and meteorological data to model the transport, weathering, and fate of spilled oil over a six-day period. Oil behavior is examined across key transformation processes, including dispersion, emulsification, evaporation, and biodegradation, using particle-based modeling and a comprehensive set of environmental inputs. The modeled results are validated against in situ observations and visual inspection data, focusing on four critical dates. The study demonstrates OpenOil’s potential for accurately simulating oil dispersion dynamics in semi-enclosed marine environments and highlights the significance of environmental forcing, vertical mixing, and shoreline interactions in determining oil fate. It concludes with recommendations for improving real-time response strategies in similar spill scenarios. Full article

This study presents a novel approach to FTIR spectroscopy at low temperatures under atmospheric pressure. The work aimed to confirm the efficiency of a fundamentally new cryogenic setup that enables material research under the specified conditions. The new technique combines a nitrogen-based cryogenic capillary cooling system with precise temperature monitoring via a PID controller, along with DRIFT spectroscopy for hydrocarbon materials. New fundamental data were obtained on the properties and behavior of hydrocarbon compounds such as methanol, kerosene, and ethanol. The IR spectra of these samples contain key characteristic vibrations of hydrocarbon functional groups, which demonstrate the effective operability of the cryogenic device. A detailed description of the setup and measurement technique is provided, along with a thorough comparison of the results with data from other authors. The application scope of the cryogenic device, the relevance of the research, and potential future developments are also discussed. Full article

Food quality and safety are essential for ensuring public health, preventing foodborne illness, reducing food waste, maintaining consumer confidence, and supporting regulatory compliance and international trade. This has led to the emergence of many research works that focus on automating and streamlining the assessment of food quality. Electronic noses have become of paramount importance in this context. We analyze the current state of research in the development of electronic noses for food quality and safety. We examined research papers published in three different scientific databases in the last decade, leading to a comprehensive review of the field. Our review found that most of the efforts use portable, low-cost electronic noses, coupled with pattern recognition algorithms, for evaluating the quality levels in certain well-defined food classes, reaching accuracies exceeding 90% in most cases. Despite these encouraging results, key challenges remain, particularly in diversifying the sensor response across complex substances, improving odor differentiation, compensating for sensor drift, and ensuring real-world reliability. These limitations indicate that a complete device mimicking the flexibility and selectivity of the human olfactory system is not yet available. To address these gaps, our review recommends solutions such as the adoption of adaptive machine learning models to reduce calibration needs and enhance drift resilience and the implementation of standardized protocols for data acquisition and model validation. We introduce benchmark comparisons and a future roadmap for electronic noses that demonstrate their potential to evolve from controlled studies to scalable industrial applications. In doing so, this review aims not only to assess the state of the field but also to support its transition toward more robust, interpretable, and field-ready electronic nose technologies. Full article

We present the validation of the latest version F Geophysical Data Records (GDR-F) for the TOPEX/POSEIDON (T/P) altimetry satellite mission. The GDR-F products represent a major evolution with respect to the preceding version B Merged Geophysical Data Records (MGDR-B) that were released more than two decades ago. Specifically, the numerical retracking of the altimeter waveforms significantly mitigates long-standing issues in the TOPEX altimeter measurements, such as drifts and hemispherical biases in the altimeter range and significant wave height. Additionally, GDR-F incorporates updated geophysical model standards consistent with current altimeter missions, improved sea state bias corrections, end-of-mission calibration for the microwave radiometer, and refined orbit ephemeris solutions. These enhancements notably decrease the variance of the Sea Surface Height Anomaly (SSHA) measurements, with along-track SSHA variance reduced by 26 cm² compared to MGDR-B and crossover SSHA variance lowered by 1 cm². GDR-F products also demonstrate improved consistency with Jason-1 measurements during their tandem mission phase, reducing the standard deviation of differences from 6 cm to 4 cm when compared to Jason-1 GDR-E data. These results confirm that GDR-F products offer a more accurate and consistent T/P data record, enhancing the quality of long-term sea level studies and supporting inter-mission altimetry continuity. Full article

Boao Jade Beach, on the east coast of Hainan Island, is a typical sandy beach and is one of the areas where typhoons frequently land in Hainan. This study examined wind speed, wind direction, and sediment transport data obtained from field meteorological stations and omnidirectional sand accumulation instruments from 2020 to 2024 to study the coastal aeolian environment and sediment transport distribution characteristics in the region. The findings provide a theoretical basis for comprehensive analyses of the evolution of coastal aeolian landforms and the evaluation and control of coastal aeolian hazards. The research results showed the following: (1) The annual average threshold wind velocity for sand movement in the study area was 6.13 m/s, and the wind speed frequency was 20.97%, mainly dominated by easterly winds (NNE, NE) and southerly winds (S). (2) The annual drift potential (DP) and resultant drift potential (RDP) of Boao Jade Belt Beach from 2020 to 2024 were 125.99 VU and 29.59 VU, respectively, indicating a low-energy wind environment. The yearly index of directional wind variability (RDP/DP) was 0.23, which is classified as a small ratio and indicates blunt bimodal wind conditions. The yearly resultant drift direction (RDD) was 329.41°, corresponding to the NNW direction, indicating that the sand on Boao Jade Belt Beach is generally transported in the southwest direction. (3) When the measured data from the sand accumulation instrument in the study area from 2020 to 2024 were used for a statistical analysis, the results showed that the total sediment transport rate in the study area was 39.97 kg/m·a, with the maximum sediment transport rate in the S direction being 17.74 kg/m·a. These results suggest that, when sand fixation systems are constructed for relevant infrastructure in the region, the direction of protective forests and other engineering measures should be perpendicular to the net direction of sand transport. Full article

Ιn this study, we evaluated the genetic resources of nine Greek sheep breeds. The genotyping data of 292 animals were acquired from Illumina’s OvineSNP50 Genotyping BeadChip. The genetic diversity and inbreeding levels were evaluated using the observed and expected heterozygosity indices, the F_IS inbreeding coefficient, and runs of homozygosity (ROH). The genetic differentiation of breeds was assessed using the F_ST index, whereas their population structure was analyzed using admixture and principal components analysis (PCA). Historical recombination patterns and genetic drift were evaluated based on linkage disequilibrium, effective population sizes, and gene flow analysis to reveal migration patterns. PCA revealed distinct clusters mostly separating mountainous, insular, and lowland breeds. The F_ST value was the lowest between Serres and Karagouniko breeds (0.050). Admixture analysis revealed a genetic substructure for Serres and Kalarritiko breeds, while Chios, followed by Katsika, demonstrated the highest within-breed genetic uniformity. ROH analysis revealed low levels of inbreeding for all breeds. Genetic introgression from both Anatolia and Eastern Europe has been evidenced for Greek sheep breeds. The results also revealed that Greek sheep breeds maintain adequate levels of genetic diversity, without signs of excessive inbreeding, and can serve as valuable resources for the conservation of local biodiversity. Full article

With the rapid advancement of Internet of Things (IoT) technology, intrusion detection systems (IDSs) have become pivotal in ensuring network security. However, the data produced by IoT devices is typically sensitive and tends to display non-independent and identically distributed (Non-IID) properties. These factors impose significant limitations on the application of traditional centralized learning. In response to these issues, this study introduces a novel IDS framework grounded in federated learning and knowledge distillation (KD), termed FD-IDS. The proposed FD-IDS aims to tackle issues related to safeguarding data privacy and distributed heterogeneity. FD-IDS employs mutual information for feature selection to enhance training efficiency. For Non-IID data scenarios, the system combines a proximal term with KD. The proximal term restricts the deviation between local and global models, while KD utilizes the global model to steer the training process of local models. Together, these mechanisms effectively alleviate the problem of model drift. Experiments conducted on both the Edge-IIoT and N-BaIoT datasets demonstrate that FD-IDS achieves promising detection performance across multiple evaluation metrics. Full article

Using a segmented linear regression model, we examined the seasonal profiles of weekly COVID-19 deaths data in Italy over a three-year-long period during which the SARS-CoV-2 Omicron and post-Omicron variants were predominant (September 2021–September 2024). Comparing the slopes of the regression segments, we were able to discuss the variation in steepness of the Italian COVID-19 mortality trend, identifying the corresponding growth/decline profile for each considered season. Our findings show that, although the COVID-19 weekly death mortality has been in a declining trend in Italy since the end of 2021 until the end of 2024, there have been increasing alterations in the COVID-19 deaths for all winters and summers of that period. These increasing mortality variations were more pronounced in winters than in summers, with an average progressive increase in the number of COVID-19 deaths, with each new week, of 55.75 and 22.90, in winters and in summers, respectively. We found that COVID-19 deaths were, instead, less frequent in the intermediate periods between winters and summers, with an average decrease of −38.01 COVID-19 deaths for each new week. Our segmented regression model has fitted well the observed COVID-19 deaths, as confirmed by the average value of the determination coefficients: 0.74, 0.63 and 0.70, respectively, for winters, summers and intermediate periods. In conclusion, favored by a general declining COVID-19 mortality trend in Italy in the period of interest, transient rises of the mortality have occurred both in winters and in summers, but received little attention because they have always been compensated by consistent downward drifts occurring during the intermediate periods between winters and summers. Full article

The observations of atmospheric, oceanic, and sea ice data from the Multidisciplinary drifting Observatory for the Study of Arctic Climate (MOSAiC) expedition were used to analyze the influence of snow redistribution and melt-pond processes on the evolution of sea ice thickness (SIT) in 2019 and 2020. To mitigate the effect of missing atmospheric observations from the time of the expedition, we used ERA5 atmospheric reanalysis along the MOSAiC drift trajectory to force the single-column sea ice model Icepack. SIT simulations from six combinations of two melt-pond schemes and three snow-redistribution configurations of Icepack were compared with observations and analyzed to investigate the sources of model–observation discrepancies. The three snow-redistribution configurations are the bulk scheme, the snwITDrdg scheme, and one simulation conducted without snow redistribution. The bulk scheme describes snow loss from level ice to leads and open water, and snwITDrdg describes wind-driven snow redistribution and compaction. The two melt-pond schemes are the TOPO scheme and the LVL scheme, which differ in the distribution of melt water. The results show that Icepack without snow redistribution simulates excessive snow–ice formation, resulting in an SIT thicker than that observed in spring. Applying snow-redistribution schemes in Icepack reduces snow–ice formation while enhancing the congelation rate. The bulk snow-redistribution scheme improves the SIT simulation for winter and spring, while the bias is large in simulations using the snwITDrdg scheme. During the summer, Icepack underestimates the sea ice surface albedo, resulting in an underestimation of SIT at the end of simulation. The simulations using the TOPO scheme are characterized by a more realistic melt-pond evolution compared to those using the LVL scheme, resulting in a smaller bias in SIT simulation. Full article

The present study analyzes the global ionospheric response to the intense geomagnetic storm of 10–11 October 2024 (SYM—H minimum of −346 nT), using observations from COSMIC—2 and Swarm satellites, GNSS TEC, and Digisondes. Significant uplift of the F-region was observed across both Hemispheres on the dayside, primarily driven by equatorward thermospheric winds and prompt penetration electric fields (PPEFs). However, this uplift did not correspond with increases in foF2 due to enhanced molecular nitrogen-promoting recombination in sunlit regions and the F2 peak rising beyond the COSMIC—2 detection range. In contrast, in the Southern Hemisphere nightside ionosphere exhibited pronounced Ne depletion and low hmF2 values, attributed to G-conditions and thermospheric composition changes caused by storm-time circulation. Strong vertical plasma drifts exceeding 100 m/s were observed during both the main and recovery phases, particularly over Ascension Island, driven initially by southward IMF—Bz-induced PPEFs and later by disturbance dynamo electric fields (DDEFs) as IMF—Bz turned northward. Swarm data revealed a poleward expansion of the Equatorial Ionization Anomaly (EIA), with more pronounced effects in the Southern Hemisphere due to seasonal and longitudinal variations in ionospheric conductivity. Additionally, the storm excited Large-Scale Travelling Ionospheric Disturbances (LSTIDs), triggered by thermospheric perturbations and electrodynamic drivers, including PPEFs and DDEFs. These disturbances, along with enhanced westward thermospheric wind and altered zonal electric fields, modulated ionospheric irregularity intensity and distribution. The emergence of anti-Sq current systems further disrupted quiet-time electrodynamics, promoting global LSTID activity. Furthermore, storm-induced equatorial plasma bubbles (EPBs) were observed over Southeast Asia, initiated by enhanced PPEFs during the main phase and suppressed during recovery, consistent with super EPB development mechanisms. Full article

This study considers methods for assessing the quality of orbital observations, quantifying drift over time, and the application of correction methods to long-term series. AIRS v6 (IR-only) satellite methane (CH₄) and carbon monoxide (CO) total column (TC) measurements were compared with NDACC ground station data from 2003 to 2022. For CH₄, negative trends were observed in the difference between satellite and ground measurements (AIRS-GR) at all 18 stations (mean drift: 1.69 × 10¹⁴ ± 0.31 × 10¹⁴ molecules/cm² per day), suggesting a shift in the orbital spectrometer parameters is probable. The application of a dynamic correction based on this drift coefficient significantly improved the correlation with satellite data for both daily means and trends at all stations. In contrast, AIRS v6 CO measurements showed a strong initial correlation (R = 0.93 for the entire dataset, and R ~ 0.8–0.95 for separate stations) without systematic drift, i.e., the trends of AIRS-GR at individual sites were oppositely directed and statistically insignificant. Therefore, the AIRS v6 CO TC satellite product does not require additional correction within this method. The developed methodology for satellite data verification and correction is supposed to be universal and applicable to other long-term orbital observations. Full article

Problem: Agriculture in India faces pressing challenges related to water scarcity, excessive pesticide use, and inefficient energy consumption, impacting both economic sustainability and environmental health. Methodology: This study integrates Life Cycle Assessment (LCA), Data Envelopment Analysis (DEA), Intelligent Management Models (IMMs), and Multi-Criteria Decision Analysis (MCDA) to assess the economic and environmental benefits of UAV-based spraying in Indian agriculture. Data were collected from UAV service providers and field trials in Punjab, Haryana, and Rajasthan. Results: UAV spraying achieved a 70% reduction in water use, 40% reduction in pesticide consumption, and a 50% reduction in CO₂ emissions compared to conventional spraying. DEA results showed higher efficiency scores for UAVs, while IMM optimization achieved 95% pesticide coverage and reduced drift by 80%. Implications: MCDA ranked government subsidies as the most effective policy intervention. These findings support UAV spraying as a viable, scalable solution for climate-smart agriculture in India, offering both productivity and sustainability gains. Full article

Camera-centric perception has matured into a cornerstone of modern autonomy, from self-driving cars and factory cobots to underwater and planetary exploration. This review synthesizes more than a decade of progress in vision-based robotic navigation through an engineering lens, charting the full pipeline from sensing to deployment. We first examine the expanding sensor palette—monocular and multi-camera rigs, stereo and RGB-D devices, LiDAR–camera hybrids, event cameras, and infrared systems—highlighting the complementary operating envelopes and the rise of learning-based depth inference. The advances in visual localization and mapping are then analyzed, contrasting sparse and dense SLAM approaches, as well as monocular, stereo, and visual–inertial formulations. Additional topics include loop closure, semantic mapping, and LiDAR–visual–inertial fusion, which enables drift-free operation in dynamic environments. Building on these foundations, we review the navigation and control strategies, spanning classical planning, reinforcement and imitation learning, hybrid topological–metric memories, and emerging visual language guidance. Application case studies—autonomous driving, industrial manipulation, autonomous underwater vehicles, planetary rovers, aerial drones, and humanoids—demonstrate how tailored sensor suites and algorithms meet domain-specific constraints. Finally, the future research trajectories are distilled: generative AI for synthetic training data and scene completion; high-density 3D perception with solid-state LiDAR and neural implicit representations; event-based vision for ultra-fast control; and human-centric autonomy in next-generation robots. By providing a unified taxonomy, a comparative analysis, and engineering guidelines, this review aims to inform researchers and practitioners designing robust, scalable, vision-driven robotic systems. Full article

To address the technical problems of broad droplet size spectrum, insufficient atomization uniformity, and spray drift in plant protection unmanned aerial vehicle (UAV) applications, this study developed a novel two-stage aerial electrostatic spraying device based on the coupled mechanisms of hydraulic atomization and electrostatic induction, and, through the integration of three-dimensional numerical simulation and additive manufacturing technology, a new two-stage inductive charging device was designed on the basis of the traditional hydrodynamic nozzle structure, and a synergistic optimization study of the charging effect and atomization characteristics was carried out systematically. With the help of a charge ratio detection system and Malvern laser particle sizer, spray pressure (0.25–0.35 MPa), charging voltage (0–16 kV), and spray height (100–1000 mm) were selected as the key parameters, and the interaction mechanism of each parameter on the droplet charge ratio (C/m) and the particle size distribution (Dv50) was analyzed through the Box–Behnken response surface experimental design. The experimental data showed that when the charge voltage was increased to 12 kV, the droplet charge-to-mass ratio reached a peak value of 1.62 mC/kg (p < 0.01), which was 83.6% higher than that of the base condition; the concentration of the particle size distribution of the charged droplets was significantly improved; charged droplets exhibited a 23.6% reduction in Dv50 (p < 0.05) within the 0–200 mm core atomization zone below the nozzle, with the coefficient of variation of volume median diameter decreasing from 28.4% to 16.7%. This study confirms that the two-stage induction structure can effectively break through the charge saturation threshold of traditional electrostatic spraying, which provides a theoretical basis and technical support for the optimal design of electrostatic spraying systems for plant protection UAVs. This technology holds broad application prospects in agricultural settings such as orchards and farmlands. It can significantly enhance the targeted deposition efficiency of pesticides, reducing drift losses and chemical usage, thereby enabling agricultural enterprises to achieve practical economic benefits, including reduced operational costs, improved pest control efficacy, and minimized environmental pollution, while generating environmental benefits. Full article