Network For Recognition Research Articles

Topology-adaptive Bayesian optimization for deep ring echo state networks in speech emotion recognition

Topology-adaptive Bayesian optimization for deep ring echo state networks in speech emotion recognition

LIMUNet: A Lightweight Neural Network for Human Activity Recognition Using Smartwatches

The rise of mobile communication, low-power chips, and the Internet of Things has made smartwatches increasingly popular. Equipped with inertial measurement units (IMUs), these devices can recognize user activities through artificial intelligence (AI) analysis of sensor data. However, most existing AI-based activity recognition algorithms require significant computational power and storage, making them unsuitable for low-power devices like smartwatches. Additionally, discrepancies between training data and real-world data often hinder model generalization and performance. To address these challenges, we propose LIMUNet and its smaller variant LIMUNet-Tiny—lightweight neural networks designed for human activity recognition on smartwatches. LIMUNet utilizes depthwise separable convolutions and residual blocks to reduce computational complexity and parameter count. It also incorporates a dual attention mechanism specifically tailored to smartwatch sensor data, improving feature extraction without sacrificing efficiency. Experiments on the PAMAP2 and LIMU datasets show that the LIMUNet improves recognition accuracy by 2.9% over leading lightweight models while reducing parameters by 88.3% and computational load by 58.4%. Compared to other state-of-the-art models, LIMUNet achieves a 9.6% increase in accuracy, with a 60% reduction in parameters and a 57.8% reduction in computational cost. LIMUNet-Tiny further reduces parameters by 75% and computational load by 80%, making it even more suitable for resource-constrained devices.

Self-cure Dual-branch Network for Facial Expression Recognition Based on Visual Sensors

Self-cure Dual-branch Network for Facial Expression Recognition Based on Visual Sensors

Multiscale attention-over-attention network for retinal disease recognition in OCT radiology images

Retinal disease recognition using Optical Coherence Tomography (OCT) images plays a pivotal role in the early diagnosis and treatment of conditions. However, the previous attempts relied on extracting single-scale features often refined by stacked layered attentions. This paper presents a novel deep learning-based Multiscale Feature Enhancement via a Dual Attention Network specifically designed for retinal disease recognition in OCT images. Our approach leverages the EfficientNetB7 backbone to extract multiscale features from OCT images, ensuring a comprehensive representation of global and local retinal structures. To further refine feature extraction, we propose a Pyramidal Attention mechanism that integrates Multi-Head Self-Attention (MHSA) with Dense Atrous Spatial Pyramid Pooling (DASPP), effectively capturing long-range dependencies and contextual information at multiple scales. Additionally, Efficient Channel Attention (ECA) and Spatial Refinement modules are introduced to enhance channel-wise and spatial feature representations, enabling precise localization of retinal abnormalities. A comprehensive ablation study confirms the progressive impact of integrated blocks and attention mechanisms that enhance overall performance. Our findings underscore the potential of advanced attention mechanisms and multiscale processing, highlighting the effectiveness of the network. Extensive experiments on two benchmark datasets demonstrate the superiority of the proposed network over existing state-of-the-art methods.

Advanced Hybrid Neural Networks for Accurate Recognition of the Extended Alphabet and Dynamic Signs in Mexican Sign Language (MSL)

The Mexican deaf community primarily uses Mexican Sign Language (MSL) for communication, but significant barriers arise when interacting with hearing individuals unfamiliar with the language. Learning MSL requires a substantial commitment of at least 18 months, which is often impractical for many hearing people. To address this gap, we present an MSL-to-Spanish translation system that facilitates communication through a spelling-based approach, enabling deaf individuals to convey any idea while simplifying the AI’s task by limiting the number of signs to be recognized. Unlike previous systems that focus exclusively on static signs for individual letters, our solution incorporates dynamic signs, such as “k”, “rr”, and “ll”, to better capture the nuances of MSL and enhance expressiveness. The proposed Hybrid Neural Network-based algorithm integrates these dynamic elements effectively, achieving an F1 score of 90.91%, precision of 91.25%, recall of 91.05%, and accuracy of 91.09% in the extended alphabet classification. These results demonstrate the system’s potential to improve accessibility and inclusivity for the Mexican deaf community.

Recognizing human activities using light-weight and effective machine learning methodologies

Background Human activity recognition (HAR) is increasingly important in enhancing healthcare systems by enabling accurate monitoring of individuals' movements through sensor data. This paper is motivated by the need to improve the accuracy of HAR, particularly for applications in e-health systems, where reliable activity detection can lead to better health outcomes. The study explores six prominent machine learning techniques—decision tree, random forest, linear regression, Naïve Bayes, k-nearest neighbor, and neural networks—to determine which methods can most effectively predict activities like walking, sitting, standing, laying, walking upstairs, and walking downstairs. Methods We employed these six machine learning algorithms to analyze a comprehensive dataset derived from various sensors. Each model was rigorously trained and evaluated to compare its effectiveness in recognizing human activities. The experiments aimed to identify strengths and weaknesses in each approach, with particular emphasis on advanced techniques such as random forest, convolutional neural networks (CNNs), and gated recurrent networks (GRNs). Results The experimental evaluation revealed that the random forest classifier, CNN, GRN, and neural networks delivered promising results, achieving high accuracy levels. Notably, the neural network model excelled, attaining an impressive accuracy of 98%. In contrast, the Naïve Bayes model did not meet the performance expectations set by the other algorithms. Conclusions This research effectively classifies activities such as sitting, standing, laying, walking, walking downstairs, and walking upstairs, underscoring the potential of machine learning in HAR. The findings highlight the superior performance of neural networks in enhancing activity recognition, which could lead to advanced applications in e-health systems and improve overall healthcare monitoring strategies.

Multivariate variational mode decomposition and 1D residual neural network for subtle feature recognition of rolling bearings

Multivariate variational mode decomposition and 1D residual neural network for subtle feature recognition of rolling bearings

Two-stage Rule-induction visual reasoning on RPMs with an application to video prediction

Raven’s Progressive Matrices (RPMs) are frequently used in evaluating human’s visual reasoning ability. Researchers have made considerable efforts in developing systems to automatically solve the RPM problem, often through a black-box end-to-end convolutional neural network for both visual recognition and logical reasoning tasks. Based on the intrinsic natures of RPM problem, we propose a Two-stage Rule-Induction Visual Reasoner (TRIVR), which consists of a perception module and a reasoning module, to tackle the challenges of real-world visual recognition and subsequent logical reasoning tasks, respectively. For the reasoning module, we further propose a “2+1” formulation that models human’s thinking in solving RPMs and significantly reduces the model complexity. It derives a reasoning rule from each RPM sample, which is not feasible for existing methods. As a result, the proposed reasoning module is capable of yielding a set of reasoning rules modeling human in solving the RPM problems. To validate the proposed method on real-world applications, an RPM-like Video Prediction (RVP) dataset is constructed, where visual reasoning is conducted on RPMs constructed using real-world video frames. Experimental results on various RPM-like datasets demonstrate that the proposed TRIVR achieves a significant and consistent performance gain compared with state-of-the-art models.

Neurons to heartbeats: spiking neural networks for electrocardiogram pattern recognition

Neurons to heartbeats: spiking neural networks for electrocardiogram pattern recognition

MSGNet: A Multi-Feature Lightweight Learning Network for Automatic Modulation Recognition

MSGNet: A Multi-Feature Lightweight Learning Network for Automatic Modulation Recognition

YOLOv8-GABNet: An Enhanced Lightweight Network for the High-Precision Recognition of Citrus Diseases and Nutrient Deficiencies

YOLOv8-GABNet: An Enhanced Lightweight Network for the High-Precision Recognition of Citrus Diseases and Nutrient Deficiencies

Multi-Scale convolutional neural network for finger vein recognition

With the continuous advancement of science and technology, an increasing number of deep learning methods are being applied in the field of finger vein recognition to describe the structural characteristics of finger veins. However, some deep learning methods fail to adequately extract longer texture features. during the feature extraction process, resulting in a decrease in the uniqueness of extracted finger vein features. Additionally, these methods tend to extract global information while neglecting the importance of local texture information. To address the aforementioned issues, this paper introduces a multiscale convolution network (MCNet) model based on finger vein structure. On one hand, a multiscale feature extraction (MFE) model based on the rectangular and square convolution kernels are employed to extract structural information from finger veins and to simultaneously enhance the features of longer texture features. On the other hand, the paper introduces a cross-information fusion attention (CFA) block that combines spatial and channel information, in order to enhance local details information and the network’s ability to extract vein patterns. The experimental results on the public datasets FV-USM, SDUMLA-HMT, and HKPU validate the effectiveness of MCNet with the recognition rates of 99.86%, 99.11%, and 99.15% respectively.

Intelligent Gear Decision Method for Vehicle Automatic Transmission System Based on Data Mining

The gear decision logic of automatic transmission directly affects the vehicle's dynamic, fuel economic, and comfort performance. This study employs data mining techniques to address the issues of low adaptability and low recognition rate in the intelligent gear decision of vehicle automatic transmission systems. The research further proposes the utilization of Kalman filter, Hidden Markov Models, and Long Short-Term Memory networks for condition feature recognition and time series classification. Subsequently, dynamic programming algorithms are employed to optimize intelligent gear decisions. Combining driver intent and driving environment, an intelligent gear decision method is formulated. The results indicate that, during a 430s driving segment, the intelligent gear decision method consumes only 464mL of fuel, closely resembling the economic strategy's 457mL, with a gear shift frequency of 53, significantly better than the 79 shifts in the economic strategy. Moreover, the error rate for slope condition recognition is only 0.062%. In a 200s coupled condition, the intelligent gear decision results in fuel consumption of 207mL, approximating the actual vehicle's 219mL, while power-shifting consumes 316mL, and economic shifting only 202mL. This study not only improves the accuracy of gear decisions but also effectively enhances vehicle operational efficiency, providing valuable insights for future automatic transmission systems with significant practical value.

Multi-source domain adaptation for EEG emotion recognition based on inter-domain sample hybridization.

Electroencephalogram (EEG) is widely used in emotion recognition due to its precision and reliability. However, the nonstationarity of EEG signals causes significant differences between individuals or sessions, making it challenging to construct a robust model. Recently, domain adaptation (DA) methods have shown excellent results in cross-subject EEG emotion recognition by aligning marginal distributions. Nevertheless, these methods do not consider emotion category labels, which can lead to label confusion during alignment. Our study aims to alleviate this problem by promoting conditional distribution alignment during domain adaptation to improve cross-subject and cross-session emotion recognition performance. This study introduces a multi-source domain adaptation common-branch network for EEG emotion recognition and proposes a novel sample hybridization method. This method enables the introduction of target domain data information by directionally hybridizing source and target domain samples without increasing the overall sample size, thereby enhancing the effectiveness of conditional distribution alignment in domain adaptation. Cross-subject and cross-session experiments were conducted on two publicly available datasets, SEED and SEED-IV, to validate the proposed model. In cross-subject emotion recognition, our method achieved an average accuracy of 90.27% on the SEED dataset, with eight out of 15 subjects attaining a recognition accuracy higher than 90%. For the SEED-IV dataset, the recognition accuracy also reached 73.21%. Additionally, in the cross-session experiment, we sequentially used two out of the three session data as source domains and the remaining session as the target domain for emotion recognition. The proposed model yielded average accuracies of 94.16 and 75.05% on the two datasets, respectively. Our proposed method aims to alleviate the difficulties of emotion recognition from the limited generalization ability of EEG features across subjects and sessions. Though adapting the multi-source domain adaptation and the sample hybridization method, the proposed method can effectively transfer the emotion-related knowledge of known subjects and achieve accurate emotion recognition on unlabeled subjects.

Automated Recognition of Submerged Body-like Objects in Sonar Images Using Convolutional Neural Networks

The Police Robot for Inspection and Mapping of Underwater Evidence (PRIME) is an uncrewed surface vehicle (USV) currently being developed for underwater search and recovery teams to assist in crime scene investigation. The USV maps underwater scenes using sidescan sonar (SSS). Test exercises use a clothed mannequin lying on the seafloor as a target object to evaluate system performance. A robust, automated method for detecting human body-shaped objects is required to maximise operational functionality. The use of a convolutional neural network (CNN) for automatic target recognition (ATR) is proposed. SSS image data acquired from four different locations during previous missions were used to build a dataset consisting of two classes, i.e., a binary classification problem. The target object class consisted of 166 196 × 196 pixel image snippets of the underwater mannequin, whereas the non-target class consisted of 13,054 examples. Due to the large class imbalance in the dataset, CNN models were trained with six different imbalance ratios. Two different pre-trained models (ResNet-50 and Xception) were compared, and trained via transfer learning. This paper presents results from the CNNs and details the training methods used. Larger datasets are shown to improve CNN performance despite class imbalance, achieving average F1 scores of 97% in image classification. Average F1 scores for target vs background classification with unseen data are only 47% but the end result is enhanced by combining multiple weak classification results in an ensemble average. The combined output, represented as a georeferenced heatmap, accurately indicates the target object location with a high detection confidence and one false positive of low confidence. The CNN approach shows improved object detection performance when compared to the currently used ATR method.

Dynamic spatial-temporal topology graph network for skeleton-based action recognition

Dynamic spatial-temporal topology graph network for skeleton-based action recognition

Integrating path signature and pen-tip trajectory features for online handwriting Yi text recognition

Recognizing online handwriting Yi text is crucial for recording and preserving Yi literature. However, the scarcity of online handwriting Yi text datasets has limited relevant research, impeding the process of Yi informatization. In this work, we use synthetic data to train models, and an Online Handwriting Yi Text Recognition Network (YTRN) is proposed, which extracts robust character features to address the gap between synthetic and real data. YTRN adeptly learns the spatial structure features from path signature feature maps and captures trajectory features from the pen-tip trajectories. Subsequently, an innovative adaptive feature fusion module integrates these two sets of features to yield more comprehensive and robust character representations. Experiments on our newly collected Yi-OLHWDB2.0 dataset demonstrate that our method outperforms previous approaches, achieving an impressive 95.67% accuracy. This highlights the model’s effectiveness in extracting comprehensive and robust features from path signature maps and pen-tip trajectories, significantly enhancing recognition accuracy and generalization.

Automated detection of stale beef from electronic nose data

AbstractAccurate detection of stale beef on the market is important for protecting the legitimate rights and interests of consumers. To this end, we combined electronic nose measurements with machine learning technology to classify beef samples. We used an electronic nose to collect information about the odor characteristics of different beef samples and used linear discriminant analysis to reduce data dimensionality. We then classified samples using the following algorithms: extreme gradient boosting, logistic regression, K‐nearest neighbor, random forest, support vector machine, and neural networks for pattern recognition. We assessed model performance using a 10‐fold cross‐validation technique. All these methods reached an accuracy of 95% or above, with F1 scores and AUC values above 0.96. The support vector machine algorithm outperformed all other models, achieving perfect recognition with 100% accuracy and F1/AUC scores of 1.0. Our study demonstrates that electronic nose data combined with support vector machine can be used to successfully discriminate between stale and fresh beef, paving the way for novel research directions in the detection of stale beef.

Field-free spin–orbit torque magnetization switching in Pt/CoTb devices grown on flexible substrates for neuromorphic computing

Flexible spintronic devices based on spin–orbit torque (SOT)-induced perpendicular magnetization switching (PMS) have attracted increasing attention due to their high storage intensity and good programming capability. However, to achieve deterministic PMS, an in-plane auxiliary magnetic field is required, which greatly limits its application. Here, we show that “robust” magnetic field-free SOT-driven PMS is realized in the oblique sputtered Pt/CoTb multilayers grown on a flexible polyimide substrate. “Robust” means the magnetic field-free SOT switching is highly repeatable and stable after 100 bending cycles under various bending conditions. Additionally, the fabricated flexible multilayers exhibit nearly linear and nonvolatile multistate plasticity as synapses and a nonlinear sigmoid activation function when acting as neurons. We construct a fully connected neural network for handwritten digit recognition, achieving an over 96.27% recognition rate. Our findings may spur further investigations on the SOT-based flexible spintronic devices for wearable artificial intelligence applications.

Detecting command injection attacks in web applications based on novel deep learning methods

Web command injection attacks pose significant security threats to web applications, leading to potential server information leakage or severe server disruption. Traditional detection methods struggle with the increasing complexity and obfuscation of these attacks, resulting in poor identification of malicious code, complicated feature extraction processes, and low detection efficiency. To address these challenges, a novel detection model, the Convolutional Channel-BiLSTM Attention (CCBA) model, is proposed, leveraging deep learning techniques to enhance the identification of web command injection attacks. The model utilizes dual CNN convolutional channels for comprehensive feature extraction and employs a BiLSTM network for bidirectional recognition of temporal features. An attention mechanism is also incorporated to assign weights to critical features, improving the model’s detection performance. Experimental results demonstrate that the CCBA model achieves 99.3% accuracy and 98.2% recall on a real-world dataset. To validate the robustness and generalization of the model, tests were conducted on two widely recognized public cybersecurity datasets, consistently achieving over 98% accuracy. Compared to existing methods, the proposed model offers a more effective solution for identifying web command injection attacks.