Search Results (546)

Deep-learning-based multiple label chest X-ray classification has achieved significant success, but existing models still have three main issues: fixed-scale convolutions fail to capture both large and small lesions, standard pooling is lacking in the lack of attention to important regions, and linear classification lacks the capacity to model complex dependency between features. To circumvent these obstacles, we propose CONVFCMAE, a lightweight yet powerful framework that is built on a backbone that is partially frozen (77.08 % of the initial layers are fixed) in order to preserve complex, multi-scale features while decreasing the number of trainable parameters. Our architecture adds (1) an intelligent global pooling module that is learnable, with $1\times 1$ convolutions that are dynamically weighted by their spatial location, and (2) a multi-head attention block that is dedicated to channel re-calibration, along with (3) a two-layer MLP that has been enhanced with ReLU, batch normalization, and dropout. This module is used to enhance the non-linearity of the feature space. To further reduce the noise associated with labels and the imbalance in class distribution inherent to the NIH ChestXray14 dataset, we utilize a combined loss that combines BCEWithLogits and Focal Loss as well as extensive data augmentation. On ChestXray14, the average ROC–AUC of CONVFCMAE is 0.852, which is 3.97 percent greater than the state of the art. Ablation experiments demonstrate the individual and collective effectiveness of each component. Grad-CAM visualizations have a superior capacity to localize the pathological regions, and this increases the interpretability of the model. Overall, CONVFCMAE provides a practical, generalizable solution to the problem of extracting features from medical images in a practical manner. Full article

With the continuous advancement of side-channel attacks (SCA), deep learning-based methods have emerged as a prominent research focus due to their powerful feature extraction and nonlinear modeling capabilities. Traditional convolutional neural networks (CNNs) excel at capturing local temporal dependencies but struggle to model long-range sequential information effectively, limiting attack efficiency and generalization. In this paper, we propose a hybrid deep neural network architecture that integrates Residual Mamba blocks with multi-layer perceptrons (MLP) to enhance the modeling of side-channel information from AES implementations. The Residual Mamba module leverages state-space modeling to capture long-range dependencies, improving the model’s global temporal perception, while the MLP module further fuses high-dimensional features. Experiments conducted on the publicly available ASCAD dataset targeting the second byte of AES demonstrate that our model achieves guessing entropy (GE) rank 1 with fewer than 100 attack traces, significantly outperforming traditional CNNs and recent Transformer-based models. The proposed approach exhibits fast convergence and high attack efficiency, offering an effective new paradigm for deep learning in side-channel analysis with important theoretical and practical implications. Full article

This study critically examines the methodological rigor in credit card fraud detection research, revealing how fundamental evaluation flaws can overshadow algorithmic sophistication. Through deliberate experimentation with improper evaluation protocols, we demonstrate that even simple models can achieve deceptively impressive results when basic methodological principles are violated. Our analysis identifies four critical issues plaguing current approaches: (1) pervasive data leakage from improper preprocessing sequences, (2) intentional vagueness in methodological reporting, (3) inadequate temporal validation for transaction data, and (4) metric manipulation through recall optimization at precision’s expense. We present a case study showing how a minimal neural network architecture with data leakage outperforms many sophisticated methods reported in literature, achieving 99.9% recall despite fundamental evaluation flaws. These findings underscore that proper evaluation methodology matters more than model complexity in fraud detection research. The study serves as a cautionary example of how methodological rigor must precede architectural sophistication, with implications for improving research practices across machine learning applications. Compared to several recent studies reporting near-perfect recall (often exceeding 99%) using complex deep models, our corrected evaluation with a simple MLP baseline yields more modest but reliable metrics, exposing the overestimation common in flawed pipelines. Full article

With the growth of the global population and the increasing scarcity of arable land, traditional agricultural production is confronted with multiple challenges, such as efficiency improvement, precision operation, and sustainable development. The progressive advancement of artificial intelligence (AI) technology has created a transformative opportunity for the intelligent upgrade of agricultural equipment. This article systematically presents recent progress in computer vision, machine learning (ML), and intelligent sensing. The key innovations are highlighted in areas such as object detection and recognition (e.g., a K-nearest neighbor (KNN) achieved 98% accuracy in distinguishing vibration signals across operation stages); autonomous navigation and path planning (e.g., a deep reinforcement learning (DRL)-optimized task planner for multi-arm harvesting robots reduced execution time by 10.7%); state perception (e.g., a multilayer perceptron (MLP) yielded 96.9% accuracy in plug seedling health classification); and precision control (e.g., an intelligent multi-module coordinated control system achieved a transplanting efficiency of 5000 plants/h). The findings reveal a deep integration of AI models with multimodal perception technologies, significantly improving the operational efficiency, resource utilization, and environmental adaptability of agricultural equipment. This integration is catalyzing the transition toward intelligent, automated, and sustainable agricultural systems. Nevertheless, intelligent agricultural equipment still faces technical challenges regarding data sample acquisition, adaptation to complex field environments, and the coordination between algorithms and hardware. Looking ahead, the convergence of digital twin (DT) technology, edge computing, and big data-driven collaborative optimization is expected to become the core of next-generation intelligent agricultural systems. These technologies have the potential to overcome current limitations in perception and decision-making, ultimately enabling intelligent management and autonomous decision-making across the entire agricultural production chain. This article aims to provide a comprehensive foundation for advancing agricultural modernization and supporting green, sustainable development. Full article

This study pioneers the integration of Sentinel-2 satellite imagery with Kolmogorov–Arnold networks (KAN) for the evaluation of chlorophyll-a (Chl-a) concentrations in inland lakes. Using Baiyangdian Lake in Hebei Province, China, as a case study, a specialized KAN architecture was designed to extract spectral features from Sentinel-2 data, and a robust algorithm was developed for Chl-a estimation. The results demonstrate that the KAN model outperformed traditional feature-engineering-based machine learning (ML) methods and standard multilayer perceptron (MLP) deep learning approaches, achieving an R² of 0.8451, with MAE and RMSE as low as 1.1920 μg/L and 1.6705 μg/L, respectively. Furthermore, attribution analysis was conducted to quantify the importance of individual features, highlighting the pivotal role of bands B3 and B5 in Chl-a retrieval. Furthermore, spatio-temporal distributions of Chl-a concentrations in Baiyangdian Lake from 2020 to 2024 were generated leveraging the KAN model, further elucidating the underlying causes of water quality changes and examining the driving factors. Compared to previous studies, the proposed approach leverages the high spatial resolution of Sentinel-2 imagery and the accuracy and interpretability of the KAN model, offering a novel framework for monitoring water quality parameters in inland lakes. These findings may guide similar research endeavors and provide valuable decision-making support for environmental agencies. Full article

Background/Objectives: Accurate diagnosis of brain tumors is one of the most important challenges in neuro-oncology since tumor classification and volumetric segmentation inform treatment planning. Two-dimensional classification and three-dimensional segmentation deep learning models can augment radiological workflows, particularly if paired with explainable AI techniques to improve model interpretability. The objective of this research was to develop a web-based brain tumor segmentation and classification diagnosis platform. Methods: A diagnosis system was developed combining 2D tumor classification and 3D volumetric segmentation. Classification employed a fine-tuned MobileNetV2 model trained on a glioma, meningioma, pituitary tumor, and normal control dataset. Segmentation employed a SegResNet model trained on BraTS multi-channel MRI with synthetic no-tumor data. A meta-classifier MLP was used for binary tumor detection from volumetric features. Explainability was offered using XRAI maps for 2D predictions and Gaussian overlays for 3D visualizations. The platform was incorporated into a web interface for clinical use. Results: MobileNetV2 2D model recorded 98.09% classification accuracy for tumor classification. 3D SegResNet obtained Dice coefficients around 68–70% for tumor segmentations. The MLP-based tumor detection module recorded 100% detection accuracy. Explainability modules could identify the area of the tumor, and saliency and overlay maps were consistent with real pathological features in both 2D and 3D. Conclusions: Deep learning diagnosis system possesses improved brain tumor classification and segmentation with interpretable outcomes by utilizing XAI techniques. Deployment as a web tool and a user-friendly interface made it suitable for clinical usage in radiology workflows. Full article

The Automatic Identification System (AIS) has become a key data source for ship behavior monitoring and maritime traffic management, widely used in trajectory prediction and anomaly detection. However, AIS data suffer from issues such as spatial sparsity, heterogeneous features, variable message formats, and irregular sampling intervals, while vessel trajectories are characterized by strong spatial–temporal dependencies. These factors pose significant challenges for efficient and accurate modeling. To address this issue, we propose a lightweight vessel trajectory prediction framework that integrates Spatial–Temporal Identity encoding with an MLP-Mixer architecture. The framework discretizes spatial and temporal features into structured IDs and uses dual MLP modules to model temporal dependencies and feature interactions without relying on convolution or attention mechanisms. Experiments on a large-scale real-world AIS dataset demonstrate that the proposed STID-Mixer achieves superior accuracy, training efficiency, and generalization capability compared to representative baseline models. The method offers a compact and deployable solution for large-scale maritime trajectory modeling. Full article

Background and Objective: Temperature management is key for reliable operation of permanent magnet synchronous motors (PMSMs). The lumped-parameter thermal network (LPTN) is fast and interpretable but struggles with nonlinear behavior under high power density. We propose OLTEM, a physics-informed deep model that combines LPTN with a thermal neural network (TNN) to improve prediction accuracy while keeping physical meaning. Methods: OLTEM embeds LPTN into a recurrent state-space formulation and learns three parameter sets: thermal conductance, inverse thermal capacitance, and power loss. Two additions are introduced: (i) a state-conditioned squeeze-and-excitation (SC-SE) attention that adapts feature weights using the current temperature state, and (ii) an enhanced power-loss sub-network that uses a deep MLP with SC-SE and non-negativity constraints. The model is trained and evaluated on the public Electric Motor Temperature dataset (Paderborn University/Kaggle). Performance is measured by mean squared error (MSE) and maximum absolute error across permanent-magnet, stator-yoke, stator-tooth, and stator-winding temperatures. Results: OLTEM tracks fast thermal transients and yields lower MSE than both the baseline TNN and a CNN–RNN model for all four components. On a held-out generalization set, MSE remains below 4.0 °C² and the maximum absolute error is about 4.3–8.2 °C. Ablation shows that removing either SC-SE or the enhanced power-loss module degrades accuracy, confirming their complementary roles. Conclusions: By combining physics with learned attention and loss modeling, OLTEM improves PMSM temperature prediction while preserving interpretability. This approach can support motor thermal design and control; future work will study transfer to other machines and further reduce short-term errors during abrupt operating changes. Full article

Recent studies on network intrusion detection using deep learning primarily focus on detecting attacks or classifying attack types, but they often overlook the challenge of attributing each attack to its specific source among many potential adversaries (multi-adversary attribution). This is a critical and underexplored issue in cybersecurity. In this study, we address the problem of attacker attribution in complex, multi-step network attack (MSNA) environments, aiming to identify the responsible attacker (e.g., IP address) for each sequence of security alerts, rather than merely detecting the presence or type of attack. We propose a deep learning approach based on Transformer encoders to classify sequences of network alerts and attribute them to specific attackers among many candidates. Our pipeline includes data preprocessing, exploratory analysis, and robust training/validation using stratified splits and 5-fold cross-validation, all applied to real-world multi-step attack datasets from capture-the-flag (CTF) competitions. We compare the Transformer-based approach with a multilayer perceptron (MLP) baseline to quantify the benefits of advanced architectures. Experiments on this challenging dataset demonstrate that our Transformer model achieves near-perfect accuracy (99.98%) and F1-scores (macro and weighted ≈ 99%) in attack attribution, significantly outperforming the MLP baseline (accuracy 80.62%, macro F1 65.05% and weighted F1 80.48%). The Transformer generalizes robustly across all attacker classes, including those with few samples, as evidenced by per-class metrics and confusion matrices. Our results show that Transformer-based models are highly effective for multi-adversary attack attribution in MSNA, a scenario not or under-addressed in the previous intrusion detection systems (IDS) literature. The adoption of advanced architectures and rigorous validation strategies is essential for reliable attribution in complex and imbalanced environments. Full article

Falls are a major safety concern in physically demanding occupations such as roofing, where workers operate on inclined surfaces under unstable postures. While inertial measurement units (IMUs) are widely used in wearable fall detection systems, they often fail to capture early indicators of instability related to foot–ground interactions. This study evaluates the effectiveness of plantar pressure sensors, alone and combined with IMUs, for fall detection on sloped surfaces. We collected data in a controlled laboratory environment using a custom-built roof mockup with incline angles of 0°, 15°, and 30°. Participants performed roofing-relevant activities, including standing, walking, stooping, kneeling, and simulated fall events. Statistical features were extracted from synchronized IMU and plantar pressure data, and multiple machine learning models were trained and evaluated, including traditional classifiers and deep learning architectures, such as MLP and CNN. Our results show that integrating plantar pressure sensors significantly improves fall detection. A CNN using just three IMUs and two plantar pressure sensors achieved the highest F1 score of 0.88, outperforming the full 17-sensor IMU setup. These findings support the use of multimodal sensor fusion for developing efficient and accurate wearable systems for fall detection and physical health monitoring. Full article

Reservoir-induced landslides threaten infrastructures and downstream communities, making continuous deformation monitoring vital. Time-series InSAR, notably the SBAS algorithm, provides high-precision surface-displacement mapping but suffers from voids due to layover/shadow effects and temporal decorrelation. Existing deep-learning approaches often operate on fixed-size patches or ignore irregular spatio-temporal dependencies, limiting their ability to recover missing pixels. With this objective, a hybrid spatio-temporal Graph Attention (ST-GAT) framework was developed and trained on SBAS-InSAR values using 24 influential features. A unified spatio-temporal graph is constructed, where each node represents a pixel at a specific acquisition time. The nodes are connected via inverse distance spatial edges to their K-nearest neighbors, and they have bidirectional temporal edges to themselves in adjacent acquisitions. The two spatial GAT layers capture terrain-driven influences, while the two temporal GAT layers model annual deformation trends. A compact MLP with per-map bias converts the fused node embeddings into normalized LOS estimates. The SBAS-InSAR results reveal LOS deformation, with 48% of missing pixels and 20% located near the Dasu dam. ST D-GAT reconstructed fully continuous spatio-temporal displacement fields, filling voids at critical sites. The model was validated and achieved an overall R² (0.907), ρ (0.947), per-map R² ≥ 0.807 with RMSE ≤ 9.99, and a ROC-AUC of 0.91. It also outperformed the six compared baseline models (IDW, KNN, RF, XGBoost, MLP, simple-NN) in both RMSE and R². By combining observed LOS values with 24 covariates in the proposed model, it delivers physically consistent gap-filling and enables continuous, high-resolution landslide monitoring in radar-challenged mountainous terrain. Full article

A chili pepper variety and origin detection system that integrates a field-programmable gate array (FPGA) with an electronic nose (e-nose) is proposed in this paper to address the issues of variety confusion and origin ambiguity in the chili pepper market. The system uses the AIRSENSE PEN3 e-nose from Germany to collect gas data from thirteen different varieties of chili peppers and two specific varieties of chili peppers originating from seven different regions. Model training is conducted via the proposed lightweight convolutional neural network ChiliPCNN. By combining the strengths of a convolutional neural network (CNN) and a multilayer perceptron (MLP), the ChiliPCNN model achieves an efficient and accurate classification process, requiring only 268 parameters for chili pepper variety identification and 244 parameters for origin tracing, with 364 floating-point operations (FLOPs) and 340 FLOPs, respectively. The experimental results demonstrate that, compared with other advanced deep learning methods, the ChiliPCNN has superior classification performance and good stability. Specifically, ChiliPCNN achieves accuracy rates of 94.62% in chili pepper variety identification and 93.41% in origin tracing tasks involving Jiaoyang No. 6, with accuracy rates reaching as high as 99.07% for Xianjiao No. 301. These results fully validate the effectiveness of the model. To further increase the detection speed of the ChiliPCNN, its acceleration circuit is designed on the Xilinx Zynq7020 FPGA from the United States and optimized via fixed-point arithmetic and loop unrolling strategies. The optimized circuit reduces the latency to 5600 ns and consumes only 1.755 W of power, significantly improving the resource utilization rate and processing speed of the model. This system not only achieves rapid and accurate chili pepper variety and origin detection but also provides an efficient and reliable intelligent agricultural management solution, which is highly important for promoting the development of agricultural automation and intelligence. Full article

The salinization of agricultural soils is a serious threat to farming and ecological balance in arid and semi-arid regions. Accurate estimation of soil water-soluble ions (calcium, carbonate, magnesium, and sulfate) is necessary for correct monitoring of soil salinization and sustainable land management. Hyperspectral ground-based data are valuable in soil salinization monitoring, but the acquisition cost is high, and the coverage is small. Therefore, this study proposes a two-stage deep learning framework with multispectral remote-sensing images. First, the wavelet transform is used to enhance the Transformer and extract fine-grained spectral features to reconstruct the ground-based hyperspectral data. A comparison of ground-based hyperspectral data shows that the reconstructed spectra match the measured data in the 450–998 nm range, with R² up to 0.98 and MSE = 0.31. This high similarity compensates for the low spectral resolution and weak feature expression of multispectral remote-sensing data. Subsequently, this enhanced spectral information was integrated and fed into a novel multiscale self-attentive Transformer model (MSATransformer) to invert four water-soluble ions. Compared with BPANN, MLP, and the standard Transformer model, our model remains robust across different spectra, achieving an R² of up to 0.95 and reducing the average relative error by more than 30%. Among them, for the strongly responsive ions magnesium and sulfate, R² reaches 0.92 and 0.95 (with RMSE of 0.13 and 0.29 g/kg, respectively). For the weakly responsive ions calcium and carbonate, R² stays above 0.80 (RMSE is below 0.40 g/kg). The MSATransformer framework provides a low-cost and high-accuracy solution to monitor soil salinization at large scales and supports precision farmland management. Full article

Wi-Fi sensing leverages variations in CSI (channel state information) to infer human activities in a contactless and low-cost manner, with growing applications in smart homes, healthcare, and security. While deep learning has advanced macro-motion sensing tasks, micro-motion sensing such as keypad stroke recognition remains underexplored due to subtle inter-class CSI variations and significant intra-class variance. These challenges make it difficult for existing deep learning models typically relying on fully connected MLPs to accurately recognize keypad inputs. To address the issue, we propose a novel approach that combines a discriminative feature extractor with a Kolmogorov–Arnold Network (KAN)-based classifier. The combined model is trained to reduce intra-class variability by clustering features around class-specific centers. The KAN classifier learns nonlinear spline functions to efficiently delineate the complex decision boundaries between different keypad inputs with fewer parameters. To validate our method, we collect a CSI dataset with low-cost Wi-Fi devices (ESP8266 and Raspberry Pi 4) in a real-world keypad sensing environment. Experimental results verify the effectiveness and practicality of our method for keypad input sensing applications in that it outperforms existing approaches in sensing accuracy while requiring fewer parameters. Full article

Accurate and real-time classification of low-latency Internet traffic is critical for applications such as video conferencing, online gaming, financial trading, and autonomous systems, where millisecond-level delays can degrade user experience. Existing methods for low-latency traffic classification, reliant on raw temporal features or static statistical analyses, fail to capture dynamic frequency patterns inherent to real-time applications. These limitations hinder accurate resource allocation in heterogeneous networks. This paper proposes a novel framework integrating wavelet transform (WT) and artificial neural networks (ANNs) to address this gap. Unlike prior works, we systematically apply WT to commonly used temporal features—such as throughput, slope, ratio, and moving averages—transforming them into frequency-domain representations. This approach reveals hidden multi-scale patterns in low-latency traffic, akin to structured noise in signal processing, which traditional time-domain analyses often overlook. These wavelet-enhanced features train a multilayer perceptron (MLP) ANN, enabling dual-domain (time–frequency) analysis. We evaluate our approach on a dataset comprising FTP, video streaming, and low-latency traffic, including mixed scenarios with up to four concurrent traffic types. Experiments demonstrate 99.56% accuracy in distinguishing low-latency traffic (e.g., video conferencing) from FTP and streaming, outperforming k-NN, CNNs, and LSTMs. Notably, our method eliminates reliance on deep packet inspection (DPI), offering ISPs a privacy-preserving and scalable solution for prioritizing time-sensitive traffic. In mixed-traffic scenarios, the model achieves 74.2–92.8% accuracy, offering ISPs a scalable solution for prioritizing time-sensitive traffic without deep packet inspection. By bridging signal processing and deep learning, this work advances efficient bandwidth allocation and enables Internet Service Providers to prioritize time-sensitive flows without deep packet inspection, improving quality of service in heterogeneous network environments. Full article