FAME: A Multi-Encoder Time Series Forecasting Model Based on Fully Actuated System Theory

Global CO2 emissions reduction requires industries to manage and understand their CO2 emission levels in real-time. This paper examines the Industrial Internet of Things (IIoT) for real-time monitoring, measurement, and visualization of reducing CO2 emissions in industrial and environmental domains. Methodology: The methodology consists of a literature review based on peer-reviewed publications and use cases to explore the current state and practical implications. Furthermore, a technical analysis of IIoT systems, CO2 sensors, and data processing techniques is also identified. Results: IIoT systems can support CO2 emission monitoring and accuracy optimization in industrial domains by combining CO2 sensors, wireless communication, and data fusion techniques. In addition, machine learning and artificial intelligence can be used to reduce anomalies in CO2 sensor readings and predictive maintenance of systems. Challenges: Challenges include interoperability, data security and system scalability. To resolve these issues standardized communication protocols, data security methods and implementation barriers should be improved. Future Directions: To enhance data processing and security features, future work should focus on integrating edge computing, artificial intelligence, machine learning, and blockchain techniques. In addition, data visualizations and cost-effective solutions should also be in focus, to provide more adoptable IIoT systems in industrial domains. Conclusion: As IIoT systems and CO2 sensor technologies evolve, IIoT systems can contribute significantly to increasing global air quality and CO2 emission control in industry, agricultural, and urban areas.

Spatio-temporal data mining is the extraction of unknown and implicit knowledge, structures, spatio-temporal relationships, or patterns not explicitly stored in spatio-temporal databases. As one of data mining techniques, forecasting is widely used to predict the unknown future based upon the patterns hidden in the current and past data. In order to achieve spatio-temporal forecasting, some mature analysis tools, e.g., time series and spatial statistics are extended to the spatial dimension and the temporal dimension respectively, or they are combined linearly as spatio-temporal integration. However, such linear combination of spatial and temporal dimensions is just a simplification of complicated spatio-temporal associations existing in complex geographical phenomena. In this study, The Support Vector Machines is introduced to construct nonlinear combination functions related to the spatial and temporal dimensions for integrated spatio-temporal forecasting. The proposed method has been tested by forecasting the annual average temperature of meteorological stations in P R China. The forecasting result shows that nonlinearly integrated spatio-temporal forecasting model via Support Vector Machines obtained better forecasting accuracy than those obtained by linear combination and other conventional methods.

Medical image segmentation is a fundamental task in medical image analysis, playing a crucial role in disease diagnosis, treatment planning, and clinical decision-making. Accurate segmentation of anatomical structures, such as blood vessels, organs, and lesions, is essential for reliable medical interpretations. However, existing segmentation models often face challenges in effectively capturing both local and global features within medical images. Traditional convolutional neural networks (CNNs) excel at extracting fine-grained local features but struggle to model long-range dependencies due to their limited receptive fields. Conversely, transformer-based models can capture global contextual relationships but often fail to preserve detailed local structures, leading to suboptimal segmentation performance. The purpose of this study is to develop SEFormer, a novel hybrid network that integrates SENet, ResNet, and Transformer to enhance medical image segmentation. SEFormer aims to effectively capture both local and global features, addressing the limitations of traditional CNNs and transformer-based models.SENet is incorporated to recalibrate feature maps by adaptively emphasizing informative features, thereby compensating for the global representation limitations of CNNs. Meanwhile, the ResNet backbone ensures deep feature extraction while maintaining computational efficiency. Additionally, the Transformer module is introduced to capture long-range dependencies and enhance contextual awareness, complementing the local feature extraction capabilities of CNNs. This hybrid approach allows SEFormer to balance fine-grained spatial details with broader contextual information, leading to more precise segmentation results.Furthermore, to mitigate information loss during feature extraction and ensure a more complete and hierarchical representation of image information, we introduce a feature pyramid structure inspired by multi-scale image pyramid models. By progressively increasing the receptive field across different scales and employing skip connections, SEFormer effectively fuses multi-scale local and global features at each stage of the pyramid. This hierarchical fusion mechanism ensures a richer and more robust feature representation, which is particularly beneficial for segmenting complex medical structures such as blood vessels. We evaluate SEFormer on the CHASE_DB1 dataset, a widely used benchmark for retinal vessel segmentation. Experimental results demonstrate that SEFormer outperforms existing state-of-the-art segmentation methods, achieving a 3.25% improvement in segmentation accuracy. Additionally, we conduct ablation studies to verify the contributions of different components within SEFormer. The results show that incorporating SENet enhances feature recalibration and channel attention, while the Transformer module significantly improves global context awareness. The feature pyramid structure further contributes to performance gains by ensuring a multi-scale representation of vascular structures. Compared to conventional CNN-based methods, our approach achieves better segmentation accuracy with improved robustness to variations in vessel thickness, noise, and image contrast. Furthermore, SEFormer maintains computational efficiency, making it suitable for real-world clinical applications where both accuracy and processing speed are critical. SEFormer provides an efficient and accurate approach for medical image segmentation by leveraging the complementary strengths of CNNs, SENet, and Transformers while integrating a feature pyramid structure to maximize feature representation. Future research directions include extending SEFormer to other medical imaging modalities, such as CT and MRI scans, as well as exploring further optimization techniques to reduce computational costs while maintaining high segmentation performance. Additionally, integrating SEFormer with active learning frameworks could further improve segmentation performance by leveraging expert-annotated data more efficiently.

The long-term prediction of hypersonic glide vehicle (HGV) trajectory based on radar tracking data still faces challenges. Existing approaches are hard to provide satisfactory performances in long-term HGV trajectory prediction tasks due to insufficient long-range dependency extraction ability and cumulative errors. To address the above issues, a trajectory prediction model physics-informed Transformer (PIT) is proposed in this work. The top$\tau$-mean attention mechanism and the generative decoder are designed in this model to extract the trajectories' long-range dependency and reduce the cumulative errors. Furthermore, the physical knowledge is applied at both the input and output ends of the model, which further improves the model's prediction accuracy even in the case of missing a small amount of radar tracking data. A simulated HGV trajectory dataset is established based on the dynamic equations for the PIT trajectory prediction model. The experimental results show that the PIT performs better than other state-of-the-art deep neural network models. The additional experiments on a small-scale dataset and datasets with missing tracking data further verify the excellent performance and robustness of the PIT in these cases.

Linear and nonlinear spectroscopy techniques are widely used to study numerous important chemical and physical processes. However, the interpretation of these experimental spectra often becomes very complicated because a particular spectrum constitutes a mere footprint of a host of various, possibly intertwined, effects. In this spirit, calculations in the time-dependent picture provide a useful tool for decoding such spectra. Nevertheless, the ultimate challenge is to devise a theoretical framework that could yield sufficient efficiency as well as accuracy to describe the molecular system of interest in a satisfactory way. The strategy is relatively straightforward for low dimensional systems that are directly tractable, e.g., with exact quantum dynamics performed on an equidistant grid. Despite the formidable overall exponential scaling, these calculations can be significantly accelerated by using higher-order split-operator propagation schemes. In general however, one is forced to seek an affordable balance between physical accuracy on one hand and computational efficiency on the other by employing, for instance, some of the techniques from the broad family of semiclassical methods based on classical trajectories. To this end, the thawed Gaussian approximation (TGA) is combined with an on-the-fly ab initio scheme (OTF-AI). The resulting OTF-AI-TGA algorithm is efficient enough to treat all vibrational degrees of freedom (DOFs) on an equal footing even in case of larger molecules such as pentathiophene (105 DOFs). Moreover, in sharp contrast to popular approaches based on global harmonic approximation, OTF-AI-TGA reproduces almost perfectly the absorption and photoelectron spectra of ammonia, i.e., spectra with strong dependence on large amplitude motions. In addition to the mere reproduction of experimental spectra, a novel systematic approach is introduced to assess the importance and the dynamical couplings of individual vibrational DOFs. This is in turn used to gain a deeper insight into the associated physical and chemical processes by attributing specific spectral features to the underlying molecular motion. Specifically, in the case of oligothiophenes, this approach was used to assign the dynamical interplay between quinoid and aromatic characters of individual rings to particular spectral patterns and, furthermore, to explain the changes in the vibrational line shape with an increasing number of rings. Furthermore, in systems that are too large to be treated with accurate quantum methods, efficient methods such as OTF-AI-TGA are expected to be useful as a preliminary tool for identification of the subspace of the important DOFs. On this subspace, one can then unleash some of the less efficient yet better-suited methods. In summary, OTF-AI-TGA combined with this novel analysis approach is intended to provide the first crucial step in a hierarchical computational protocol for studying large molecules such as dyes.

EMBANet: A flexible efficient multi-branch attention network.

The present dissertation describes the research work carried out towards: I. developing of analytical and numerical models for accurate and efficient analysis of laminates and sandwiches under different loading conditions (static analyses, impulsive loading and low velocity impacts); II. developing of techniques to improve structural performances of composites structures by tuning their energy absorption mechanism; A higher order theory (AD-ZZ) including zig-zag effects, transverse shear continuity, variable transverse displacement and layerwise representation is presented for analysis of laminates and sandwiches. This theory is cast in such a way that the unknowns in the model will involve only five variables like equivalent single layer models: the three displacement components and the two shear rotations of the normals on the reference surface. It is advantageous to formulate the theory in this way, as it provides the opportunity to refine its through-the-thickness representation by adding computational layers in the thickness direction without increasing the number of unknowns. After extensive evaluation of the accuracy obtained by the above mentioned theory for a variety of structures, the AD-ZZ model is enriched by adding a set of continuity functions in order to treat laminates and structures with in-plane discontinuities (e.g. bonded joints, two material wedge) under a unified approach without adding new unknowns. Speaking of techniques to improve structural performances, two approaches are presented: variable stiffness composites and stitching. The former tool works on the strain energy of the structure computing fibre distributions that minimize unwanted energy contributions and maximize the wanted ones. The latter prescribes the insertions of transverse reinforcements in sandwich structures. The homogenized mechanical properties of this structure, required by the AD-ZZ model to perform the analysis, are evaluated through virtual material tests using a 3D FE model. The results obtained applying these techniques demonstrate that both are effective in reducing transverse stresses at critical interfaces and, in certain cases, in improving bending stiffness. New C0 displacement-based and stress-based finite elements for analysis of laminated and sandwich plates based on the AD-ZZ model are developed. To achieve this goal, a new technique that converts derivatives of functional degrees of freedom contained in the AD-ZZ is presented. Both the elements satisfy all the requirements of computationally efficient finite element models for analysis of multilayered structures, namely 1) the number of degrees of freedom is independent from the number of layers; 2) no shear correction factors or penalty number are added in the formulation. Consistent shear fields are obtained for the present finite element formulations. Numerical results demonstrate that these new elements are robust, accurate and computationally efficient for analysis of multilayered structures. In order to study structures subjected to impact loading a new simulation procedure is developed. It is based on the AD-ZZ model and considers the crushing behaviour of soft-like media without performing each time a detailed 3D finite element analysis. Numerical results show that the response of composites undergoing low velocity impact is very accurately predicted with low computational effort. Overall, the present procedure holds great promise for analysis of laminated and sandwich structures undergoing low velocity impact

Mass spectrometry (MS) analysis plays a crucial role in the biomedical field; however, the high dimensionality and complexity of MS data pose significant challenges for feature extraction and classification. Deep learning has become a dominant approach in data analysis, and while some deep learning methods have achieved progress in MS classification, their feature representation capabilities remain limited. Most existing methods rely on single-channel representations, which struggle to effectively capture structural information within MS data. To address these limitations, we propose a Multi-Channel Embedding Representation Module (MSMCE), which focuses on modeling inter-channel dependencies to generate multi-channel representations of raw MS data. Additionally, we implement a feature fusion mechanism by concatenating the initial encoded representation with the multi-channel embeddings along the channel dimension, significantly enhancing the classification performance of subsequent models. Experimental results on four public datasets demonstrate that the proposed MSMCE module not only achieves substantial improvements in classification performance but also enhances computational efficiency and training stability, highlighting its effectiveness in raw MS data classification and its potential for robust application across diverse datasets.

Local to non-local: Multi-scale progressive attention network for image restoration

Sensorimotor optimization in higher dimensions

During the last decade, the deluge of multimedia data has impacted a wide range of research areas, including multimedia retrieval, 3D tracking, database management, data mining, machine learning, social media analysis, medical imaging, and so on. Machine learning is largely involved in multimedia applications of building models for classification and regression tasks, etc., and the learning principle consists in designing the models based on the information contained in the multimedia dataset. While many paradigms exist and are widely used in the context of machine learning, most of them suffer from the `curse of dimensionality', which means that some strange phenomena appears when data are represented in a high-dimensional space. Given the high dimensionality and the high complexity of multimedia data, it is important to investigate new machine learning algorithms to facilitate multimedia data analysis. To deal with the impact of high dimensionality, an intuitive way is to reduce the dimensionality. On the other hand, some researchers devoted themselves to designing some effective learning schemes for high-dimensional data. In this survey, we cover feature transformation, feature selection and feature encoding, three approaches fighting the consequences of the curse of dimensionality. Next, we briefly introduce some recent progress of effective learning algorithms. Finally, promising future trends on multimedia learning are envisaged.

A quantum system of a particle interacting with a (non-Abelian) gauge field on the non-simply connected domain M=R2\{an}Nn=1 is considered, where an, n=1,...,N, are fixed isolated points in R2. The gauge potential A of the gauge field is a p×p anti-Hermitian matrix-valued 1-form on M, and may be strongly singular at the points an, n=1,...,N. If A is flat, then the physical momentum and the position operators {Pj,qj}2j=1 of the particle satisfy the canonical commutation relations (CCR) with two degrees of freedom on a suitable dense domain of the Hilbert space L2(R2;Cp). A necessary and sufficient condition for this representation to be the Schrödinger 2-system is given in terms of the Wilson loops of the rectangles not intersecting an, n=1,...,N. This also gives a characterization for the representation to be non-Schrödinger. It is proven that, for a class of gauge potentials, which is not necessarily flat, Pj is essentially self-adjoint. Moreover, an example, which gives a class of non-Schrödinger representations of the CCR with two degrees of freedom, is discussed in some detail.

The rapid proliferation of deepfake technology, driven by advanced generative models like GANs and Diffusion Models, has enabled the creation of hyper-realistic manipulated media that threatens information integrity, personal reputation, and national security. As these tools become more accessible, the need for automated, robust, and efficient detection systems is critical. Existing detection methodologies largely rely on Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), or Vision Transformers (ViTs). However, each approach faces significant limitations. CNN-based methods suffer from "temporal blindness," failing to detect inconsistencies across video frames. Hybrid CNN-RNN models address this but are hindered by slow training speeds and limited capacity for modeling long-range dependencies. While Transformer-based models excel at capturing global context, their quadratic computational complexity (O(N2)) makes them computationally expensive and memory-intensive, rendering them inefficient for processing long, high-resolution video sequences. Furthermore, many current models struggle to generalize effectively to unseen datasets and novel manipulation techniques. To overcome these challenges, this paper proposes a novel hybrid deepfake detection framework that integrates an EfficientNet backbone with the Mamba Selective State Space Model (SSM). This architecture leverages EfficientNet for robust spatial feature extraction and utilizes Mamba for temporal sequence modeling. Mamba introduces a content-aware selection mechanism and operates with linear-time complexity (O(N2)), allowing it to efficiently capture subtle long-range temporal inconsistencies that traditional models miss, without the heavy computational cost of Transformers. Experimental results demonstrate that this proposed system offers a scalable, efficient solution with superior generalization capabilities, making it highly suitable for real-time deepfake detection in diverse and complex video environments. Keywords: Deepfake Detection, GAN, Selective State Space Models (SSM), Mamba model, Vision Mamba (Vim), Temporal Sequence Modeling, EfficientNet Backbone, machine learning, deep learning, convolutional neural networks, transformers, attention mechanisms, benchmarking systems, datasets, Synthetic Media Detection, Linear-Time Complexity, Video Anomaly Detection, Computational Efficiency, Hybrid CNN-SSM, Facial Forgery Detection, Long-Range Dependency Modeling.

MPEG video traffic is expected to represent most of the load in the future high-speed networks. Adequate traffic models for MPEG Variable Bit-Rate (VBR) video are important for network design, performance evaluation, admission control and resource allocation.It is well-known that VBR video exhibits long-range correlations over arbitrarily long time-scales, a phenomenon usually referred to as Long-Range Dependence (LRD).Many models for VBR video traffic have been proposed in the literature. However, while the correlation between Groups of Pictures (GOPs) has been widely analyzed in literature, little effort has been devoted up to now to the generation of the different frame-types within a MPEG GOP.Due to the difficulty and complexity involved in this issue, this is often neglected even though it is a fundamental characteristic of MPEG traffic and it might have an important impact on queueing performance. In fact, many works avoid to descend to the frame level proposing models for the GOP process.In this paper, we analyze the impact on queueing performance of different solutions proposed in the literature for generating the different types of MPEG frames within a GOP (these solutions have never been compared and analyzed jointly). In this way, we can be in a good position to use the best model for MPEG in terms of simplicity, computational efficiency and queueing performance, depending on the performance metric to study (loss rate, mean delay and jitter).

As the resolution and data volume of remote sensing imagery continue to grow, achieving efficient compression without sacrificing reconstruction quality remains a major challenge, given that traditional handcrafted codecs often fail to balance rate-distortion performance and computational complexity, while deep learning-based approaches offer superior representational capacity. However, challenges remain in achieving a balance between fine-detail adaptation and computational efficiency. Mamba, a state–space model (SSM)-based architecture, offers linear-time complexity and excels at capturing long-range dependencies in sequences. It has been adopted in remote sensing compression tasks to model long-distance dependencies between pixels. However, despite its effectiveness in global context aggregation, Mamba’s uniform bidirectional scanning is insufficient for capturing high-frequency structures such as edges and textures. Moreover, existing visual state–space (VSS) models built upon Mamba typically treat all channels equally and lack mechanisms to dynamically focus on semantically salient spatial regions. To address these issues, we present an innovative architecture for distant sensing image compression, called the Multi-scale Channel Global Mamba Network (MGMNet). MGMNet integrates a spatial–channel dynamic weighting mechanism into the Mamba architecture, enhancing global semantic modeling while selectively emphasizing informative features. It comprises two key modules. The Wavelet Transform-guided Local Structure Decoupling (WTLS) module applies multi-scale wavelet decomposition to disentangle and separately encode low- and high-frequency components, enabling efficient parallel modeling of global contours and local textures. The Channel–Global Information Modeling (CGIM) module enhances conventional VSS by introducing a dual-path attention strategy that reweights spatial and channel information, improving the modeling of long-range dependencies and edge structures. We conducted extensive evaluations on three distinct remote sensing datasets to assess the MGMNet. The results of the investigations revealed that MGMNet outperforms the current SOTA models across various performance metrics.