Real-Time Anomaly Detection in High-Velocity Data Streams Using Hybrid Deep Learning Architectures

Magnetic resonance imaging (MRI), a widely adopted modality in clinical practice, enables the acquisition of multi-contrast images from the same anatomical structure, commonly referred to as multimodal images. Integrating these diverse modalities is crucial for enhancing model performance across a variety of medical image analysis tasks. However, in real-world clinical scenarios, it is often impractical to acquire all MRI modalities simultaneously due to factors such as patient discomfort, time constraints, and scanning costs. As a result, synthesizing missing modalities from available ones has emerged as an effective solution. To address these challenges, we propose HMF-MambaINR, a hierarchical multi-scale feature fusion network for cross-modality MRI synthesis. The model integrates Mamba-based Selective State Space Modeling (SSM) and implicit neural representation (INR) to capture long-range dependencies and enable continuous spatial reconstruction. A Multi-Feature Extraction Block (MFEB) captures local and global representations via multi-scale receptive fields, while a Modulation Fusion Module (MFM) adaptively fuses multi-modal features with dynamic weighting. Extensive experiments show that HMF-MambaINR surpasses state-of-the-art CNN-, Transformer-, and Mamba-based methods in synthesizing missing MRI modalities. Notably, the synthesized MRI images received positive feedback from radiologists in terms of image quality, contrast, and structural contour accuracy, highlighting the potential of the proposed method as a practical tool for clinical applications.

Capturing complex temporal dependencies across multiple scales remains a fundamental challenge in time series forecasting. Transformer-based models have achieved impressive performance on sequence tasks, but vanilla designs often struggle to integrate information from both local fluctuations and global trends, especially in non-stationary sequences. We propose iTransformer-FFC, a novel forecasting framework that addresses these issues through frequency-domain analysis and multi-scale feature fusion. In particular, iTransformer-FFC introduces a Fast Fourier Convolution (FFC) module to transform time series data into the frequency domain, isolating dominant periodic components and attenuating noise before attention is applied. A hierarchical feature fusion mechanism that integrates features extracted at multiple temporal resolutions then jointly models global and local temporal patterns, while a factorized self-attention architecture reduces the quadratic complexity of standard Transformers, improving efficiency while maintaining accuracy. Together, these innovations enable more effective long-range dependency modeling and adaptability to regime shifts in the data. Extensive experiments on five public benchmark datasets demonstrate that iTransformer-FFC consistently outperforms state-of-the-art models, including the vanilla Transformer, an earlier iTransformer variant, and PatchTST. Notably, our model achieves on average an 8.73% lower MSE and 6.95% lower MAE than the best performing baseline, confirming its superior predictive accuracy and generalization in multi-scale time series forecasting through its innovative integration of frequency-domain analysis, hierarchical feature fusion, and factorized attention mechanisms.

Peptide drugs have become an important direction in new drug development due to their high specificity, designability, and good biocompatibility. However, toxicity risk remains a key bottleneck limiting the translation of peptides from lead peptides to clinical applications. Therefore, there is an urgent need to develop a high-precision and scalable toxicity assessment framework. Based on the above understanding, we propose a dual-modal feature fusion framework, PTP-SMGCA, for accurate peptide toxicity discrimination. It captures local motifs and long-range dependencies through sequence pathways and characterizes atomic-level topology and bonding semantics through molecular graph pathways. By using bidirectional cross-attention to achieve dynamic alignment of molecular graph features and sequence features, consistency evidence can be automatically screened between modalities and modality-specific noise can be suppressed. Experimental results show that PTP-SMGCA significantly improves the performance of peptide toxicity prediction, with an AUROC of 0.9289 and an AUPRC of 0.9397. On the same data set, its predictive performance is better than that of several other advanced tools, including ATSE, CAPTP, ToxiPep, Toxinpred2 (AAC-based RF), and Toxinpred2 (Hybrid). Through interpretability analysis, PTP-SMGCA could effectively identify key amino acids and the key groups they form. Attention scores are read out, revealing that graph features and sequence features dynamically aligned by bidirectional cross-attention played a crucial role in our model's prediction of peptide toxicity. In summary, this peptide toxicity prediction framework enables rapid screening of toxic peptides, providing a new approach for the development of therapeutic peptide drugs.

The ML-based inferential statistics framework presents a novel solution for database query optimization that addresses critical challenges in statistics maintenance and cardinality estimation. By combining Bayesian learning and reinforcement learning modules, the framework enables continuous adaptation to changing data patterns while minimizing computational overhead. The solution offers improved query performance through better plan selection, reduced resource consumption, and enhanced accuracy in cardinality estimation. The framework's dynamic histogram redistribution mechanism ensures optimal statistics maintenance in high-throughput environments, making it particularly effective for enterprise-scale databases with rapidly evolving data distributions.

Time series forecasting plays a crucial role in industrial control and related domains. However, existing models still face challenges in balancing representation capacity and computational efficiency when handling complex data characterized by high dimensionality, non-stationarity, and long-range dependencies. To address these issues, this paper proposes FAME (Fully Actuated system theory-based Multi-Encoder model), a novel time series forecasting framework that introduces fully actuated system theory into temporal prediction for the first time. The model jointly models temporal and channel dimensions as two degrees of freedom, and designs an innovative DFG-MLP (Dynamic Feature Gated MLP) to enhance traditional MLP architectures. Experimental results demonstrate that FAME achieves 16.1% average MSE reduction and 11.4 % MAE improvement compared to mainstream models (Autoformer, FEDformer, etc.) in long-term forecasting tasks, while maintaining significantly lower computational overhead. The proposed model validates the effectiveness and engineering practicality of Fully Actuated system theory in temporal forecasting

Transferring knowledge from one domain to another is challenging due to a number of reasons. Since both conditional and marginal distribution of the training data and test data are non-identical, model trained in one domain, when directly applied to a different domain, is usually low in accuracy. For many applications with large feature sets, such as text document, sequence data, medical data, image data of different resolutions, etc. two domains usually do not contain exactly the same features, thus introducing large numbers of when considered over the union of features from both domains. In other words, its marginal distributions are at most overlapping. In the same time, these problems are usually high dimensional, such as, several thousands of features. Thus, the combination of high dimensionality and missing values make the relationship in conditional probabilities between two domains hard to measure and model. To address these challenges, we propose a framework that first brings the marginal distributions of two domains closer by filling up those missing values of disjoint features. Afterwards, it looks for those comparable sub-structures in the latent-space as mapped from the expanded feature vector, where both marginal and conditional distribution are similar. With these sub-structures in latent space, the proposed approach then find common concepts that are transferable across domains with high probability. During prediction, unlabeled instances are treated as queries, the mostly related labeled instances from out-domain are retrieved, and the classification is made by weighted voting using retrieved out-domain examples. We formally show that importing feature values across domains and latent semantic index can jointly make the distributions of two related domains easier to measure than in original feature space, the nearest neighbor method employed to retrieve related out domain examples is bounded in error when predicting in-domain examples. Software and datasets are available for download.

Network connectivity is a fundamental requirement for intervehicle communication and services in VANETs. The fast movement of the vehicles results in rapid changes in network topology generating dynamic variation in network connectivity, which causes dynamic variation in network connectivity. In this paper, we analyze the evolution of the VANETs connectivity based on the realistic data collected by the GPS devices installed on taxis in Shanghai. The study of the evolution characteristics shows that the number of connected components, the size of the giant component, and the connectivity length time series possess strong temporal correlation and long-range dependence. As a result, the short-range dependent processes such as the exponential, normal, or Poisson distributions cannot capture the statistical distribution characteristics. Moreover, all connectivity parameters cannot be described by unimodal distributions. We also conclude that although the whole topology of the VANETs is broken into a large number of small-sized connected components under small transmission radius, the size of the giant component keeps stable when transmission radius reaches the threshold. These results will be helpful to optimize the RSUs deployment and transmission radius to keep the giant component stable and design more efficient strategies to forward the information in VANETs.

Research on the dynamic linkage among the carbon emission trading, energy and capital markets

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.

Summary1. Fish community structure and habitat distribution of the abundant species roach, perch and ruffe were studied in Lake Nordborg (Denmark) before (August 2006) and after (August 2007) aluminium treatment to reduce internal phosphorus loading.2. Rapid changes in fish community structure, abundance and habitat distribution occurred following a decline in in‐lake phosphorus concentrations from 280 to 37 μg P L−1 and an increase in Secchi depth transparency from 1.1 to 1.9 m (August). The proportion of perch in overnight gill net catches increased, whilst roach decreased, and the average weight of all key species increased.3. The habitat distribution of perch and roach changed from a high proportion in the upper pelagic and littoral zones in 2006, towards enhanced proportions in the deeper pelagic and profundal zone in 2007. The abundance of large‐bodied zooplankton increased and the abundance of benthic invertebrates decreased in the same period, suggesting that the habitat shift was not induced by food limitation.4. Ruffe shifted from the littoral and upper profundal zones towards the deep profundal zone, likely reflecting an increased predation risk in the littoral zone and better oxygen conditions in the deep profundal.5. Our results indicate that enhanced risk of predation in the upper pelagic and the littoral zones and perhaps improved oxygen concentrations in the deeper profundal zone at decreasing turbidity are responsible for the observed habitat shift. The results indicate that fish respond rapidly to changes in nutrient state, both in terms of community structure and habitat use.

Time series anomaly detection is an important research topic in the field of intelligent operation and maintenance. When software systems are frequently updated with continuous integration and deployment, the distribution of KPI data will also change, and the accuracy of anomaly detection models will inevitably decrease. To tackle this problem, we propose an active anomaly detection framework named Active-MTSAD suitable for multi-dimensional time series, combining unsupervised anomaly detection and active learning. The active learning module introduces three feedback strategies, namely denominator penalty, negative penalty, and metric learning, to learn new anomalous patterns under new data distribution. In metric learning, we consider the difference between normal and abnormal samples in reconstruction error and latent space. We conduct extensive experiments on a large-scale public dataset and a real-world dataset coming from Tencent. The experimental results show that Active-MTSAD can still achieve excellent performance in real scenarios where the distribution changes with only 0.2% of labels.

Blood flow (BF) increases with increasing exercise intensity in skeletal, respiratory, and cardiac muscle. In humans during maximal exercise intensities, 85% to 90% of total cardiac output is distributed to skeletal and cardiac muscle. During exercise BF increases modestly and heterogeneously to brain and decreases in gastrointestinal, reproductive, and renal tissues and shows little to no change in skin. If the duration of exercise is sufficient to increase body/core temperature, skin BF is also increased in humans. Because blood pressure changes little during exercise, changes in distribution of BF with incremental exercise result from changes in vascular conductance. These changes in distribution of BF throughout the body contribute to decreases in mixed venous oxygen content, serve to supply adequate oxygen to the active skeletal muscles, and support metabolism of other tissues while maintaining homeostasis. This review discusses the response of the peripheral circulation of humans to acute and chronic dynamic exercise and mechanisms responsible for these responses. This is accomplished in the context of leading the reader on a tour through the peripheral circulation during dynamic exercise. During this tour, we consider what is known about how each vascular bed controls BF during exercise and how these control mechanisms are modified by chronic physical activity/exercise training. The tour ends by comparing responses of the systemic circulation to those of the pulmonary circulation relative to the effects of exercise on the regional distribution of BF and mechanisms responsible for control of resistance/conductance in the systemic and pulmonary circulations. © 2012 American Physiological Society. Compr Physiol 2:321‐447, 2012.

Medical time series analysis, particularly for electrocardiogram (ECG) and electroencephalogram (EEG) signals, is essential in modern diagnostics, supporting early detection of conditions such as arrhythmias and epileptic seizures. However, existing approaches often struggle to capture multi-scale periodic patterns and longrange dependencies while meeting real-time processing demands. The envisioned 6G networks, with their terahertz communication and integrated sensing and communication (ISAC) capabilities, will generate vast volumes of high-fidelity physiological data at the network edge. This paradigm shift intensifies the conflict between the computational complexity of advanced AI models and the limited resources of edge devices, creating a critical bottleneck for deploying sophisticated analytics in real-world healthcare scenarios. To overcome these limitations, this paper introduces a 6G-enabled hierarchical contrastive learning framework, referred to as Hierarchical Contrastive Learning for Multi-Scale Medical time series analysis (HCL-MSM), which integrates three core components: a signal-adaptive encoder based on multi-period decomposition and 2D convolution, a patient-level contrastive module enhanced with decomposable multi-scale mixing, and a 6G-edge deployment module optimized via quantization and pruning. The framework effectively models nested physiological rhythms and cross-time dependencies in medical data, while maintaining low-latency operation under resource-constrained edge environments. We evaluated HCL-MSM on multiple clinical datasets under simulated 6G settings. Our framework achieves significant gains in arrhythmia detection, seizure prediction, and neurological monitoring.We evaluated HCL-MSM on multiple clinical datasets under simulated 6G settings. Our framework achieves significant gains in arrhythmia detection (F1-score: 86.39 percent), seizure prediction (Recall: 87.72 percent), and neurological monitoring (Recall: 87.8 percent), outperforming existing state-of- the-art methods.

Constructing subsurface models that accurately reproduce geological heterogeneity and their associated uncertainty is critical to many geoscience and engineering applications. For hydrocarbon reservoir modeling and forecasting, for example, spatial variability must be consistent with geological processes, geophysical measurements, and time records of fluid production measurements. Generating such subsurface models can be time-consuming; conditioning them to different types of measurements is even more computationally intensive and technically challenging. Using too many free variables can cause overfitting of the data, thereby decreasing the predictive ability of the model; high dimensionality also slows convergence during history matching of fluid production measurements. To contend with these problems, we introduce here a new machine learning approach, referred to as the stochastic pix2pix method, which parameterizes high-dimensional, stochastic reservoir models into low-dimensional Gaussian random variables in latent space. Many of the world’s significant hydrocarbon fields originate from fluvial or turbidite deposits, with sedimentary processes and spatial distribution of rock facies significantly influencing their flow behavior. Rule- or object-based methods are commonly used to model geostatistically these types of reservoirs. Here, we introduce a new and efficient machine learning-based reservoir modeling workflow capable of generating 2D fluvial reservoir models that account for the available field data and the geometries of different facies. By constraining subsurface model realizations to available geophysical and petrophysical interpretations, multiphysics inversion can be greatly accelerated, thereby requiring only production history matching. Although our models are 2D, an extension to 3D can be readily implemented through zone-by-zone (i.e., reservoir unit) modeling and conditioning. The proposed method and workflow also partially solve the common problem of machine learning methods, wherein mapping low- to high-resolution images often yield reduced spatial variability. The proposed method is an extension of conditional generative adversarial networks, in which we incorporate a novel penalty term into the loss function in order to generate various realizations honoring the same conditional data, such as structural interpretation from seismic data, and borehole measurements at key well locations. We “train” the generative model on geologically realistic, multivariate spatial models generated with a rule-based fluvial reservoir simulator. Each high-resolution training image includes five lithofacies with distinct petrophysical property trends. To evaluate the performance of the proposed method, visual inspection, indicator variograms and multiple point density function are applied to gauge how well the realizations reproduce existing patterns in the true model. A new metric, the mean categorical error, is proposed to quantify how well the realizations match the conditioning data. The proposed method correctly reproduces patterns even when the conditioning data are significantly different from those in the training set. Likewise, the method can perform continuous model modifications, meaning that the machine learning procedure effectively reproduces the migration rule of a meandering (fluvial) system.

This study proposes an adaptive spatiotemporal graph contrastive learning model to address the challenges of anomaly detection in cloud service systems under highly dynamic topologies, complex dependencies, and multi-source heterogeneous monitoring data. The model constructs a dynamic spatiotemporal dependency graph to achieve adaptive modeling of structural relationships among service nodes and incorporates temporal encoding to capture multi-scale temporal variations, thereby achieving joint spatiotemporal feature representation. It consists of three main components: adaptive graph construction, spatiotemporal feature encoding, and contrastive learning optimization, which together enhance anomaly identification in an unsupervised manner through feature aggregation and discrimination. First, a learnable adjacency matrix is generated based on the dynamic correlations among service metrics to reflect the evolution of service dependencies. Then, structural and temporal features are jointly extracted through the integration of graph convolution and recurrent neural units, forming representations that are globally consistent and locally sensitive. Finally, an InfoNCE-based contrastive loss function is introduced to strengthen the aggregation of normal samples and the separation of anomalous ones in the latent space, improving model robustness and generalization. Experiments conducted on real cloud monitoring data include multi-dimensional comparisons and sensitivity analyses covering key factors such as time window length, attention head number, node failure ratio, and anomaly sample ratio. The results show that the proposed method outperforms existing baseline models in accuracy, precision, recall, and F1-Score, maintaining high detection accuracy and stable performance in complex cloud environments and providing an efficient and robust solution for anomaly identification and intelligent operation in large-scale distributed systems.