Milena Radenković

Expert Systems with Applications

Whale identification and size estimation in satellite imagery via intelligent subtle perception

Siqi Wang, Baoxiang Huang, Milena Radenkovic, Ge Chen

https://www.sciencedirect.com/science/article/pii/S0957417425043933

Protecting whales is vital for preserving ecological balance because they are essential to marine ecosystems. Conventional detection techniques, such as acoustic technology and visual observation, are expensive, ineffective, and susceptible to environmental influences. Although satellite remote sensing provides a more comprehensive view of whale monitoring, difficulties still exist due to the complexity of the marine environment and the scarcity of datasets. Deep learning, fortunately, is becoming a powerful tool that can accelerate the optimization of subtle perception for a variety of multidisciplinary applications that enable high-precision and accurate real-time ocean observation. Here, a comprehensive methodology is proposed to implement whale detection and size estimation in satellite images, leveraging advanced artificial intelligence and data augmentation strategies to overcome the challenges. First, several techniques, including grayscale processing, Gaussian noise addition, random flipping and cropping, histogram equalization, image overlay, and Clustering Generative Adversarial Networks, are employed to generate synthetic images to augment the whale satellite dataset. Second, a custom detection model is extended with TripletAttention (TA) modules for accurate feature extraction and detection performance in complex marine environments. The expanded dataset is then used to train the model, achieving an mAP0.5 score of 0.89, indicating high accuracy in whale detection. Next, the spatial resolution of the images is used to estimate the size of the discovered whales, providing valuable data for biological studies. Finally, the proposed method is extended to monitor whale activity in specific regions, such as Hawaii, confirming peak activity levels from December to April. This research supports the effective application of artificial intelligence in the detection of large marine species.

Whale identification; Satellite remote sensing technology; Data augmentation; WST-YOLOv9 network; Subtle perception

Časopis

Journal of Sensor and Actuator Networks

Next Generation Intelligent Mobile Edge Networks for Improving Service Provisioning in Indonesian Festivals

Ayu, Vittalis, and Milena Radenkovic

https://www.mdpi.com/2224-2708/15/1/19

first_page settings Order Article Reprints Open AccessArticle Next Generation Intelligent Mobile Edge Networks for Improving Service Provisioning in Indonesian Festivals by Vittalis Ayu * [ORCID] and Milena Radenkovic [ORCID] School of Computer Science, The University of Nottingham, Nottingham NG8 1BB, UK * Author to whom correspondence should be addressed. J. Sens. Actuator Netw. 2026, 15(1), 19; https://doi.org/10.3390/jsan15010019 Submission received: 15 December 2025 / Revised: 1 February 2026 / Accepted: 2 February 2026 / Published: 6 February 2026 (This article belongs to the Section Communications and Networking) Download keyboard_arrow_down Browse Figures Versions Notes Abstract Indonesia is a country of vast geographical and cultural diversity, hosting numerous cultural festivals annually, such as Sekaten, Labuhan, and the Lembah Baliem Festival. However, as the world’s largest archipelago country, Indonesia faces geographical challenges in terms of ensuring the reliability of communication networks, particularly in maintaining user experience in high-density, short-duration traffic burst environments, such as festivals. The nation’s network connectivity relies heavily on satellite networks and Palapa Ring, a national fibre-optic backbone network that comprises a combination of inland and underwater networks, connecting major and remote islands to the global internet. Although this solution can provide a baseline for broadband connectivity, an adaptive intelligent mobile edge-based solution is needed to complement the existing network infrastructure in order to meet the dynamic demands of localised and transient traffic surges across multiple temporary, geographically dispersed festival sites in both urban and rural areas. In this paper, we present a multimodal study that combines network connectivity measurements during a festival with an extensive user analysis of festival participants and organisers to investigate reliability gaps in user experience regarding network connectivity. Our findings show that internet connectivity was intermittently disrupted during the festival, and our user analysis revealed a gap between customer expectations and perceptions of network service quality and the provision of application services in a heterogeneous festival environment. To address this challenge, we propose a novel next-generation intelligent festival mobile edge framework, MobiFest, which integrates the multi-layer Cognitive Cache which has geospatial–temporal edge intelligence for localised service provisioning to improve the delivery of application services in both urban and rural festival environments. In our extensive experiments, we employ smart garbage as our use case and demonstrate how our complex, multimodal intelligent network protocol SmartGarbiC, designed based on MobiFest for garbage management services, outperforms state-of-the-art and benchmark protocols.

Indonesia’s festival; mobile intelligent edge; opportunistic networks; application services

Časopis

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing

Optical Remote Sensing Image Super-Resolution via Adaptive High-Pass Filtering and Window Attention Mechanism

Qikai Zhou, Baoxiang Huang, He Gao, Milena Radenkovic, Ge Chen

https://ieeexplore.ieee.org/document/11535408

Remote sensing image super-resolution (SR) aims to enhance image resolution, critical for various applications. Conventional convolutional neural networks (CNNs) often struggle to recover fine textures and sharp edges, and are limited in adaptively combining local and global features. To address this, we propose Adaptive High-Pass Spatial Gate Transformer Network (AHGTNet), a novel cascade network for optical remote sensing image SR. AHGTNet integrates an Adaptive High-Pass Enhancement Block (AHEB) and a Residual Gate Window Attention Blocks (RGWB). The AHEB dynamically extracts and emphasizes high-frequency features, crucial for detailed textures and edges, by generating adaptive filtering weights and applying grouped high-pass operations. Subsequently, the RGWB efficiently models both local details and long-range dependencies, facilitating a comprehensive integration of contextual features. Extensive experiments on three optical remote sensing datasets demonstrate AHGTNet's effectiveness. Compared with state-of-the-art methods, our model consistently achieves competitive, and often leading, reconstruction quality in terms of PSNR and SSIM. Furthermore, AHGTNet shows favorable performance on real-world SR tasks, adeptly recovering detailed textures while preserving overall scene coherence.

Remote sensing, Super-resolution , Adaptive high-pass filter , Window attention mechanism

Časopis

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing

Mesoscale Westward Eddy Trajectory Prediction With Memory Augmented Neural Network

Wanchuan Kan, Baoxiang Huang, Milena Radenkovic, Xinmin Zhang, Ge Chen

https://ieeexplore.ieee.org/document/10759624

Accurate prediction of oceanic eddy trajectory is crucial for monitoring ocean climate change, but the complex dynamics mechanism and changeable environmental effects make it difficult. In recent years, many deep-learning methods have been proposed to solve this problem. However, the complexity of high-dimensional models increases the prediction accuracy as well as calculation cost. In this article, a parsimonious and interpretable network with external memory of the Rossby wave is constructed to implement westward mesoscale eddy trajectory prediction. Specifically, 1) fundamental multilayer perceptrons are utilized to extract cross-variable features, and gate recurrent units with fewer gates are employed to capture temporal corrections; 2) an external memory unit to retain the phase speed of long Rossby wave across different scales is designed to maintain simplicity and efficiency within the network; 3) the network structure includes an external memory module responsible for reading the phase speed of long Rossby wave from the external memory unit; and 4) this information is then interacted with Rossby wave related features of the eddy and corrected the output of prediction module to enhance forecasting outcomes. Experiments on dataset benchmarks demonstrate the effectiveness of the proposed method. Our method outperforms the baseline methods in terms of accuracy and computational cost, with mean geodesic distance errors of 7.52 km for three-day prediction while taking lower computational cost and training time.

Eddy trajectory prediction , external memory , gate recurrent unit (GRU) , multilayer perceptron (MLP) , Rossby wave

Časopis

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing

Long-Term Forecasting of Long-Lived Oceanic Eddies Using an Improved Informer Framework

Yimin Wang, Baoxiang Huang, Milena Radenkovic, Wanchuan Kan, Xinmin Zhang, Ge Chen

https://ieeexplore.ieee.org/document/11146589

Long-term prediction of long-lived mesoscale oceanic eddies plays a crucial role in ocean dynamics research, as its accuracy impacts climate change assessments, ocean energy transport analysis, and marine resource management. However, due to the nonlinear, multiscale nature of vortex dynamics, it remains challenging. To address this, we propose an informer-based deep learning framework enhanced with two key innovations: 1) an adaptive feature selection module that dynamically prioritizes input features to improve prediction robustness and 2) a geospatial loss function based on the Haversine formula to optimize spatial accuracy. In addition, we investigate the unique influence of dipole eddies on trajectory predictability and analyzes how different eddy lifecycle stages affect prediction performance, offering theoretical and empirical insights into these complex patterns. Experiments show that our method outperforms seven baseline models across multiple metrics, achieving a mean geodesic distance of 13.624 km in the representative 30-d input and 15-d prediction setting and reducing error accumulation in long-term forecasts. Specifically, our work focuses on eddies with lifespans exceeding one year, providing dedicated predictive tools tailored for these long-lived ocean phenomena. These advances provide a foundation for future integration of physical constraints and environmental variables, promising enhanced predictive tools for ocean dynamics research.

Deep learning , dipole eddy interaction , feature selection , long-term trajectory prediction , ocean dynamics

Časopis

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing

ARU^2-Net: A Deep Learning Approach for Global-Scale Oceanic Eddy Detection

Junmin Geng, He Gao, Baoxiang Huang, Milena Radenkovic, Ge Chen

https://ieeexplore.ieee.org/document/10572225

Ocean eddies have a significant impact on marine ecosystems and the climate because they transport essential substances in the ocean. Detection of ocean eddies has become one of the most active topics in physical ocean research. In recent years, research based on deep learning has mainly focused on regional oceans, with small and specific data and relatively general detection results. This study processes the global eddy by pixel-by-pixel classification and generates a global eddy classification map with a resolution of 720 x 1440, which expands the data volume and improves the generality of the data. Moreover, a high-precision attention residual U^2-Net model, referred to as ARU^2-Net, is proposed, which is suitable for mining eddy surface features from sea level anomaly (SLA) and sea surface temperature (SST) data in the global ocean. ARU^2-Net for eddies, and helps ARU^2-Net to better identify the eddy categories. Finally, we demonstrate the effectiveness of our approach on the global eddy dataset, achieving a test performance of 94.926%, significantly exceeding previous detection in some areas.

ARU ^2 -Net architecture , attention mechanism , deep learning , global ocean eddy , intelligent detection

Časopis

IEEE Transactions on Geoscience and Remote Sensing

Intelligent Sparse2Dense Profile Reconstruction for Predicting Global Subsurface Chlorophyll Maxima

Yongjun Yu, Baoxiang Huang, Milena Radenkovic, Tingting Wang, Ge Chen

https://ieeexplore.ieee.org/document/10684819

Subsurface chlorophyll maxima (SCM) is a crucial ecological indicator for marine ecosystems. Previous studies have indicated that this phenomenon is globally widespread. Although the biogeochemical Argo assimilation results have yielded positive results, the sparse data prevents them from being effectively used in oceanographic operations. Considering the dependence of ocean parameter, a deep learning model termed AT-GRU based on gated recurrent units is proposed. By incorporating an attention mechanism, the model can effectively address missing data in the biogeochemical-Argo (BGC-Argo) profiles, achieving the transition of data from sparse to dense (Sparse2Dense) and improving the accuracy of estimating subsurface chlorophyll-a (Chla) concentration. Specifically, the dataset of satellite remote sensing data and the associated BGC-Argo profiles is first established. AT-GRU is employed to reconstruct Chla concentration profiles from 1 to 300 m, utilizing several sources of ocean surface data. Next, an in-depth investigation is conducted to determine the characteristics of SCM. The objective is to enable wider research on SCM by analyzing vertical Chla profiles in four geographical locations. Finally, the general improvement of skill performance metrics, with R-squared reaching 0.84, demonstrates the feasibility of the proposed methodology through extensive experiments. In addition, we apply AT-GRU to global surface satellite data from January 2023 and compare the results with numerical modeling data to further validate the performance. This study presents promising opportunities for leveraging artificial intelligence in subsurface oceanic phenomena with the idea of Sparse2Dense and holds significant implications for the field of marine ecology.

Bidirectional gated recurrent unit (Bi-GRU) , biogeochemical-Argo (BGC-Argo) , remote sensing , residual connection module , self-attention , sparse to dense (Sparse2Dense) , subsurface chlorophyll maxima

Časopis

IEEE Transactions on Geoscience and Remote Sensing

Global Oceanic Mesoscale Eddies Trajectories Prediction With Knowledge-Fused Neural Network

Xinmin Zhang, Baoxiang Huang, Ge Chen, Linyao Ge, Milena Radenkovic, Guojia Hou

https://ieeexplore.ieee.org/document/10497608

Efficient eddy trajectory prediction driven by multiinformation fusion can facilitate the scientific research of oceanography, while the complicated dynamics mechanism makes this issue challenging. Benefiting from ocean observing technology, the eddy trajectory dataset can be qualified for data-intensive research paradigms. In this article, the dynamics mechanism is used to inspire the design idea of the eddy trajectory prediction neural network (termed EddyTPNet) and is also transformed into prior knowledge to guide the learning process. This study is among the first to implement eddy trajectory prediction with physics informed neural network. First, an in-depth analysis of the kinematic characteristics indicates that the longitude and latitude of the trajectory should be decoupled; second, the directional dispersion prior knowledge of global eddy propagation is embedded into the decoder of the EddyTPNet to improve the performance; finally, EddyTPNet predicts global eddy trajectories through pretraining and adapts to complex local regions via model transfer. Extensive experimental results demonstrate that EddyTPNet can reliably forecast the motion of eddies for the next seven days, ensuring a low daily mean geodetic error. This exploratory study provides valuable insights into solving the prediction problem of ocean phenomena by using knowledge-based time-series neural networks.

Deep learning , directional divergence physical information , eddy trajectory prediction , knowledge-fused neural network

Časopis

Remote Sensing of Environment

Deep learning solver unites SDGSAT-1 observations and Navier–Stokes theory for oceanic vortex streets

Gao H., Huang B., Chen G., Xia L., Radenkovic M.

https://www.sciencedirect.com/science/article/pii/S0034425724004516

The world’s first scientific satellite for sustainable development goals (SDGSAT-1) provides valuable data about offshore small-scale ocean phenomena, including the K?rm?n vortex street phenomenon. Although the simulation of the oceanic vortex street phenomenon is crucial for understanding not only the mechanisms of vortex formation in fluid dynamics but also their impact on the surrounding environment, the traditional simulation relies on the strong theoretical hypothesis of Navier–Stokes equations. Here, we propose a self-supervised neural network with high generalization ability to implement Navier–Stokes equations, simulating realistic oceanic vortex streets. Specifically, the physical informed convolutional neural network is first employed to determine the corresponding pressure and velocity fields, achieving accurate simulation of oceanic vortex streets with lower computational cost; Then, the observational islands in SDGSAT-1 imagery are embedded as obstacles, meanwhile, the marine background field including wind and terrain is synchronously incorporated to achieve more realistic simulation results compared with traditional methods; Finally, the morphological parameters of oceanic vortex streets are calculated and associated analysis are carried out to deepen our understanding of small scale vortex street phenomena. In addition, the experimental results demonstrated our proposed method can obtain promising time efficiency. With this partial differential equation deep learning solver framework combining observation and theory, there will be potential to expedite the cognitive process of oceanic phenomena.

SDGSAT-1K?rm?n vortex streetNavier–Stokes equationsDeep learning

Časopis

Intelligent Computing

Mitigating Cache Pollution Attack Using Deep Learning in Named Data Networking (NDN)

Hamdi, Mohd Maizan Fishol; Chen, Zhiyuan; Radenkovic, Milena

https://link.springer.com/chapter/10.1007/978-3-031-62277-9_28

With the explosive growth of Internet traffic, Internet communication pays more attention to the data itself rather than where it is physically located. To better cope with Internet usage, a revolutionized shift from host-centric end-to-end communication to receiver-driven content retrieval through Information-Centric Networking (ICN) with Named Data Networking (NDN) has emerged as a big player in this new paradigm. New capabilities such as caching everywhere have made it more challenging to protect all caches from new breed of security issues especially Cache Pollution Attack (CPA). This paper presents a summary of Caching security challenges in NDN but focuses mainly on CPA. The adaptation of Machine Learning (ML) based technology is a promising emerging and proven in experimental-based simulation using ndnSIM. We show that our Python-based DL algorithm can improve the detection of CPA and outperforms some state-of-the-art techniques that mostly works on numerical counters and probability-based algorithm.

Kategorije objave: casopis

Časopis

IEEE Geoscience and Remote Sensing Letters

Oceanic Eddy Identification Using Pyramid Split Attention U-Net With Remote Sensing Imagery

Nan Zhao, Baoxiang Huang, Jie Yang, Milena Radenkovic, Ge Chen

https://ieeexplore.ieee.org/document/10041953

Oceanic eddy is the ubiquitous ocean flow phenomenon, which has been the key factor in the transportation of ocean energy and materials. Consequently, oceanographic understanding can be enhanced by the intelligent identification of eddy. State-of-the-art deep learning technologies are gradually improving identification methods. This letter proposes the pyramid split attention (PSA) eddy detection U-Net architecture (PSA-EDUNet) that targets oceanic eddy identification from ocean remote sensing imagery. As for the PSA-EDUNet, its inspiration comes from U-Net, which contains encoder and decoder parts, making the integration of inferior and senior features efficient and ensuring the feature information will not be lost in large quantities through nonlinear connection mode. Meanwhile, the PAS module is introduced to enhance feature extraction. In terms of the fusion data, the sea surface feature is the main criterion of eddy identification, including sea surface temperature (SST) and sea level anomaly (SLA). The experiments are implemented on the Kuroshio Extension (KE) and the South Atlantic regions, the results demonstrate that the proposed method can outperform other methods, especially for eddy edges and small-scale eddies.

Deep learning , oceanic eddy identification , pyramid split attention (PSA) , U-Net network

Časopis

Science of the Total Environment

Instant deep sea debris detection for maneuverable underwater machines to build sustainable ocean using deep neural network

B Huang, G Chen, H Zhang, G Hou, M Radenkovic

https://www.sciencedirect.com/science/article/pii/S0048969723014420

Deep sea debris is any persistent man-made material that ends up in the deep sea. The scale and rapidly increasing amount of sea debris are endangering the health of the ocean. So, many marine communities are struggling for the objective of a clean, healthy, resilient, safe, and sustainably harvested ocean. That includes deep sea debris removal with maneuverable underwater machines. Previous studies have demonstrated that deep learning methods can successfully extract features from seabed images or videos, and are capable of identifying and detecting debris to facilitate debris collection. In this paper, the lightweight neural network (termed DSDebrisNet), which can leverage the detection speed and identification performance to achieve instant detection with high accuracy, is proposed to implement compound-scaled deep sea debris detection. In DSDebrisNet, a hybrid loss function considering the illumination and detection problem was also introduced to improve performance. In addition, the DSDebris dataset is constructed by extracting images and video frames from the JAMSTEC dataset and labeled using a graphical image annotation tool. The experiments are implemented on the deep sea debris dataset, and the results indicate that the proposed methodology can achieve promising detection accuracy in real-time. The in-depth study also provides significant evidence for the successful extension branch of artificial intelligence to the deep sea research domain.

Deep sea debris; Deep learning-based detection method; Marine pollution; Seafloor

Časopis

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing

WildFishNet: Open Set Wild Fish Recognition Deep Neural Network With Fusion Activation Pattern

Xiaoya Zhang, Baoxiang Huang, Ge Chen, Milena Radenkovic, Guojia Hou

https://ieeexplore.ieee.org/document/10197176

Wild fish recognition is a fundamental problem of ocean ecology research and contributes to the understanding of biodiversity. Given the huge number of wild fish species and unrecognized category, the essence of the problem is an open set fine-grained recognition. Moreover, the unrestricted marine environment makes the problem even more challenging. Deep learning has been demonstrated as a powerful paradigm in image classification tasks. In this article, the wild fish recognition deep neural network (termed WildFishNet) is proposed. Specifically, an open set fine-grained recognition neural network with a fused activation pattern is constructed to implement wild fish recognition. First, three different reciprocal inverted residual structural modules are combined by neural structure search to obtain the best feature extraction performance for fine-grained recognition; next, a new fusion activation pattern of softmax and openmax functions is designed to improve the recognition ability of open set. Then, the experiments are implemented on the WildFish dataset that consists of 54 459 unconstrained images, which includes 685 known classes and 1 open set unrecognized category. Finally, the experimental results are analyzed comprehensively to demonstrate the effectiveness of the proposed method. The in-depth study also shows that artificial intelligence can empower marine ecosystem research.

Deep neural network , fusion activation pattern , neural structure search , open set fine-grained recognition , wild fish recognition

Časopis

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing

SCMNet: Toward Subsurface Chlorophyll Maxima Prediction Using Embeddings and Bi-GRU Network

Ao Wang, Baoxiang Huang, Jie Yang, Ge Chen, Milena Radenkovic

https://ieeexplore.ieee.org/document/10288355

A subsurface chlorophyll maximum is an important ecological feature of planktonic ecosystems. Although the vertical profiles can be determined through the implementation of biogeochemical (BGC)-Argo buoy, this method is not compatible with the ocean observation requirements of high-resolution spatiotemporal measurements. Here, we demonstrate that deep learning can proficiently fill in these observational gaps when combined with sea surface data from ocean scolor radiometry. First, the sparse vertical profile data of BGC-Argo is fused with sea surface data to construct the benchmark dataset for deep learning. Second, encouraged by the idea of dense numerical representations, the comprehensive model combined with coupled embedding and bidirectional gated recurrent unit is proposed to inverse the vertical profile with BGC-Argo and satellite data. Then, the in-depth spatiotemporal analysis of the subsurface chlorophyll maxima phenomenon is performed by the parametric equation method and deep learning method as well. Finally, extensive experiments in the Northwest Pacific were conducted to demonstrate the effectiveness of the proposed methodology. The impressive results indicate that the proposed method can compensate for the lack of sparse in situ observations of chlorophyll concentration, the determination coefficient is increased by more than 20%. This study is of great significance to marine ecology and provides important insight into artificial intelligence in the study of subsurface oceanic phenomena.s

Biogeochemical (BGC)-Argo , bidirectional gated recurrent unit (bi-GRU) network , deep embedding , remote sensing , subsurface chlorophyll maxima

Časopis

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing

A Spectral Sequence-Based Nonlocal Long Short-Term Memory Network for Hyperspectral Image Classification

Baoxiang Huang, Zhipu Wang, Jie Shang, Ge Chen, Milena Radenkovic

https://ieeexplore.ieee.org/document/9736454

Efficient classification for hyperspectral image (HSI), which assigns each pixel of the image into a specific category, has been a critical research topic in the HSI analysis area. Under the supervised classification settings, the deep learning approaches are very useful for label prediction. However, most deep learning modeling methods cannot get the utmost out of spectral information, which is critically important for object interpretation. Consequently, a spectral sequence-based nonlocal long short-term memory (LSTM) network for HSI classification is proposed in this article. To boost the dominant role of spectral information, the LSTM network for sequential data processing is employed. Furthermore, the nonlocal diverse regions are exploited to learn contextual features for stronger discriminative ability. Meanwhile, the attention mechanism is adopted to increase the classification performance. Experiments on Salinas, Indian Pines, Pavia University, and Houston University datasets are implemented. Benefiting from the idea of processing spectral bands as sequence data and nonlocal diverse regions, spectral-spatial sssinformation is fully fused to achieve accurate pixel-level classification. It can be demonstrated that the proposed method can achieve better quantitative and qualitative classification results compared with the other state-of-the-art classification methods.

Attention mechanism , hyperspectral image (HSI) classification , long short-term memory (LSTM) , nonlocal diverse region , spectral information

Časopis

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing

An Efficient Oceanic Eddy Identification Method With XBT Data Using Transformer

Hongfeng Zhang, Baoxiang Huang, Ge Chen, Linyao Ge, Milena Radenkovic

https://ieeexplore.ieee.org/document/9944871

Oceanic mesoscale eddies are relatively small, short-lived circulation patterns that are approximately in geostrophic balance. Meanwhile, eddies are omnipresent and can be characterized by dynamic sea level anomalies and temperature anomalies. This makes the eddy identification mainstream with Sea Level Anomaly (SLA). Unfortunately, nearly 90% of sea level dynamic anomalies caused by oceanic eddies cannot be observed due to insufficient resolution of satellite altimeters. Combining in situ Expendable Bathythermograph (XBT) profiles data, and sea surface temperature data calibrated by the altimeter, this article proposes a deep neural network to identify subsurface oceanic eddies and inverse the corresponding sea surface eddy properties. First, the eddies identified by SLA are purified to match the corresponding vertical profile dataset. Then, a neural network with a self-attention mechanism is constructed by combining the eddy vertical profile structure with temporal and spatial characteristics and external features to effectively identify the eddy. Furthermore, the eddy properties including radius, amplitude, and energy are inversion with XBT profile and SST features. Finally, the experimental results show that the accuracy of eddy classification can reach 98.22%, which demonstrates that vertical profiles can be used to classify eddies effectively. Subsequent reclassification of the outside altimeter-identified eddies recaptured about 36% of the eddies. The authenticity of the newly identified eddies can be demonstrated by statistical model validation as well as validation of sea surface temperature anomaly. These results indicate that the subsurface eddies identification can be implemented by vertical profiles with deep learning

Artificial intelligence , deep learning , deep neural network , oceanic eddy , vertical structure , XBT profile

Časopis

Deep Sea Research Part I: Oceanographic Research Papers,

Deep Blue AI: A New Bridge from Data to Knowledge for the Ocean Science

Chen, G.; Huang, B.; Chen, X.; Ge, L.; Radenkovic, M.; Ma, Y.

https://www.sciencedirect.com/science/article/pii/S0967063722001984

The global ocean is the largest ecosystem on our planet, the scientific understanding of which is the first step towards the sustainable development of the perpetual ocean. With the continuous advance of observation technology, ocean science has entered the era of big data which significantly facilitates its junction with artificial intelligence (AI). This paper aims to answer the question “can AI serve as a new bridge from oceanic data to knowledge?” by summarizing a large number of investigations and implementing experiments of application examples. First, the AI frameworks of features engineering, classification, and detection, as well as time series forecasting that explores the emerging AI oceanography, are systematically deliberated; Then several case studies with different depths of the ocean are presented to demonstrate the effectiveness of these new data-to-knowledge bridges, including ocean current reconstruction, Arctic sea ice prediction, oceanic eddy identification with profiling data, and deep-sea debris detection for oceanic ecology; Finally, recommendations to strengthen the ocean AI in several typical application areas are proposed. Consensus has been reached that big data science is one of the most dynamic branches in interdisciplinary fields, of which deep blue AI (DBAI) might be one of the most universal methodologies. Consequently, the potential of DBAI needs to be widely and urgently explored and revealed. Taking advantage of exponentially increasing big ocean data, it is highly anticipated that many oceanography problems will be successfully solved by using AI-based knowledge discovery methodologies, through which the DBAI technology itself will be significantly advanced.

Perpetual ocean; Big ocean data; Artificial intelligence; Knowledge discovery; Feature engineering; Classification and detection; Time series forecasting

Časopis

Journal of Sensor and Actuator Networks

Resistance to Cybersecurity Attacks in a Novel Network for Autonomous Vehicles

Brocklehurst, C.; Radenkovic, M.

https://www.mdpi.com/2224-2708/11/3/35

The increased interest in autonomous vehicles has led to the development of novel networking protocols in VANETs In such a widespread safety-critical application, security is paramount to the implementation of the networks. We view new autonomous vehicle edge networks as opportunistic networks that bridge the gap between fully distributed vehicular networks based on short-range vehicle-to-vehicle communication and cellular-based infrastructure for centralized solutions. Experiments are conducted using opportunistic networking protocols to provide data to autonomous trams and buses in a smart city. Attacking vehicles enter the city aiming to disrupt the network to cause harm to the general public. In the experiments the number of vehicles and the attack length is altered to investigate the impact on the network and vehicles. Considering different measures of success as well as computation expense, measurements are taken from all nodes in the network across different lengths of attack. The data gathered from each node allow exploration into how different attacks impact metrics including the delivery probability of a message, the time taken to deliver and the computation expense to each node. The novel multidimensional analysis including geospatial elements provides evidence that the state-of-the-art MaxProp algorithm outperforms the benchmark as well as other, more complex routing protocols in most of the categories. Upon the introduction of attacking nodes however, PRoPHET provides the most reliable delivery probability when under attack. Two different attack methods (black and grey holes) are used to disrupt the flow of messages throughout the network and the more basic protocols show that they are less consistent. In some metrics, the PRoPHET algorithm performs better when under attack due to the benefit of reduced network traffic.

VANETS; opportunistic networks; security

Časopis

Frontiers in Marine Science

Deep blue artificial intelligence for knowledge discovery of the intermediate ocean

Chen Ge , Huang Baoxiang , Yang Jie , Radenkovic Milena , Ge Linyao , Cao Chuanchuan , Chen Xiaoyan , Xia Linghui , Han Guiyan , Ma Ying

https://www.frontiersin.org/journals/marine-science/articles/10.3389/fmars.2022.1034188

Oceans at a depth ranging from ∼100-m to ∼1,000-m (defined as the intermediate water here), though poorly understood compared to the sea surface, is a critical layer of the Earth system where many important oceanographic processes take place. Advances in ocean observation and computer technology have allowed ocean science to enter the era of big data (to be precise, big data for the surface layer, small data for the bottom layer, while the intermediate layer sits in between), and greatly promoted our understanding of near-surface ocean phenomena. During the past few decades, however, the intermediate ocean is also undergoing profound changes as a result of global warming, the research and prediction of which are of intensive concern. Due to the lack of three-dimensional ocean theories and field observations, how to remotely sense the intermediate ocean from space becomes a very attractive but challenging scientific issue. With the rapid development of the next generation of information technology, Artificial Intelligence (AI) has built a new bridge from data science to marine science (called Deep Blue AI, abbreviated as DBAI), which acts as a powerful weapon to extend the paradigm of modern oceanography in the era of Metaverse. This review first introduces the basic prior knowledge of water movement in the ∼100-m ocean and vertical stratification within the ∼1,000-m depths, as well as the data resources provided by satellite remote sensing, field observation, and model reanalysis for DBAI. Then, three universal DBAI methodologies, namely the associative statistical, physically informed, and mathematically driven neural networks, are elucidated in the context of intermediate ocean remote sensing. Finally, the unique advantages and potentials of DBAI in data mining and knowledge discovery are demonstrated in a top-down way of “surface-to-interior” via several typical examples in physical and biological oceanography.

Kategorije objave: Casopis

Časopis

Pattern Recognition

Nonlocal graph theory based transductive learning for hyperspectral image classification

Baoxiang Huang, Linyao Ge, Ge Chen, Milena Radenkovic, Xiaopeng Wang, Jinming Duan, Zhenkuan Pan

https://doi.org/10.1016/j.patcog.2021.107967

Hyperspectral Image classification plays an important role in the maintenance of remote image analysis, which has been attracting a lot of research interest. Although various approaches, including unsupervised and supervised methods, have been proposed, obtaining a satisfactory classification result is still a challenge. In this paper, an efficient transductive learning method using variational nonlocal graph theory for hyperspectral image classification is proposed. First, the nonlocal vector neighborhood similarity is employed to build sparse graph representation. Then the variational segmentation framework is extended to label space, and the vectorization nonlocal energy function is constructed. Next, a fast comprehensive alternating minimization iteration algorithm is designed to implement labels transductive learning. At the same time, the labeled sample constraints are doubled ensured with simplex projection. Finally, experiments on six widely used hyperspectral image datasets are implemented, compared with other state-of-the-art classification methods, the classification results demonstrate that the proposed method has higher classification performance. Benefiting from graph theory and transductive idea, the proposed classification method can propagate labels and overcome the very high dimensionality and limited labeling problem to some extent.

Transductive learning; Nonlocal graph; Label propagation; Variational method; Alternating direction method of multipliers; Hyperspectral image classification

Časopis