Recognition Model Research Articles

Feature-Enhanced Multimodal Interaction model for emotion recognition in conversation

Feature-Enhanced Multimodal Interaction model for emotion recognition in conversation

A Novel Dual-Task Model for EEG-Based Emotion and Cognition Recognition

A Novel Dual-Task Model for EEG-Based Emotion and Cognition Recognition

A discriminative model for scale, translation and rotation invariant face recognition.

A discriminative model for scale, translation and rotation invariant face recognition.

Enhanced Entity Recognition of Islamic Hadiths based-on Hybrid LSTM and AraBERT Model

This paper focuses on the training, evaluation and development of named entity recognition (NER) models designed for Islamic hadiths in Arabic Utilizing the Hadith Noor dataset, the study uses the BIO (Basic, In, Out) tagging scheme to classify words or tokens in NER tasks and the segmentation of the text into individual tokens. The right-skewed distribution revealed by examining the lengths of the Islamic hadiths revealed a right-skewed distribution, indicating that shorter texts are more common. Texts less than 100 words were most prevalent, followed by texts between 100 and 200 words, while texts longer than 200 words were rare. The dataset identifies eight types of entities, such as common names among narrators and locations. The study by training the three models AraBERT, LSTM and the hybrid model AraBERT-LSTM on Arabic text processing respectively, the hybrid model showed a performance, efficiency and accuracy of 0.981, outperforming the rest of the models, confirming its worth and reliability in NER tasks for natural language in Arabic, especially Islamic hadiths, which opens the way for exploring further investigations for future research in natural language processing.

Graphical User Interface for Handwritten Mathematical Expression Employing RNN-based Encoder-decoder Model

Aim: Scientific, technical, and educational research domains all heavily rely on handwritten mathematical expressions. The extensive use of online handwritten mathematical expression recognition is a consequence of the availability of strong computational touchscreen appliances, such as the recent development of deep neural networks as superior sequence recognition models. Background: Further investigation and enhancement of these technologies are vital to tackle the contemporary obstacles presented by the widespread adoption of remote learning and work arrangements as a result of the global health crisis. Objective: Handwritten document processing has gained more attention in the last ten years due to notable developments in deep neural network-based computer vision models and sequence recognition, as well as the widespread proliferation of touch and pen-enabled smartphones and tablets. It comes naturally to people to write by hand in daily interactions. Method: In this patent article, authors implemented Hand written expressions using RNNbased encoder for the CROHME dataset. Later, the proposed model was validated using CNNbased encoder and End-to-end encoder decoder techniques. The proposed model is also validated on other datasets. Results: The RNN-based encoder model yields 82.78%, while the CNN-based encoder model and end-to-end encoder-decoder technique yield 81.38% and 80.73%, respectively. Conclusion: 1.6% accuracy improvement was attained over CNN-based encoder while 2.4% accuracy improvement over end-to-end encoder-decoder. CROHME dataset 2019 version results in better accuracy than other datasets.

GSE: A global-local storage enhanced video object recognition model.

The presence of substantial similarities and redundant information within video data limits the performance of video object recognition models. To address this issue, a Global-Local Storage Enhanced video object recognition model (GSE) is proposed in this paper. Firstly, the model incorporates a two-stage dynamic multi-frame aggregation module to aggregate shallow frame features. This module aggregates features in batches from each input video using feature extraction, dynamic multi-frame aggregation, and centralized concatenations, significantly reducing the model's computational burden while retaining key information. In addition, a Global-Local Storage (GS) module is constructed to retain and utilize the information in the frame sequence effectively. This module classifies features using a temporal difference threshold method and employs a processing approach of inheritance, storage, and output to filter and retain features. By integrating global, local and key features, the model can accurately capture important temporal features when facing complex video scenes. Subsequently, a Cascaded Multi-head Attention (CMA) mechanism is designed. The multi-head cascade structure in this mechanism progressively focuses on object features and explores the correlations between key and global, local features. The differential step attention calculation is used to ensure computational efficiency. Finally, we optimize the model structure and adjust parameters, and verify the GSE model performance through comprehensive experiments. Experimental results on the ImageNet 2015 and NPS-Drones datasets demonstrate that the GSE model achieves the highest mAP of 0.8352 and 0.8617, respectively. Compared with other models, the GSE model achieves a commendable balance across metrics such as precision, efficiency, and power consumption.

Wireless communication interference signal recognition model based on deep CNN

Wireless communication interference signal recognition model based on deep CNN

Air target intent recognition method combining graphing time series and diffusion models

Air target intent recognition holds significant importance in aiding commanders to assess battlefield situations and secure a competitive edge in decision-making. Progress in this domain has been hindered by challenges posed by imbalanced battlefield data and the limited robustness of traditional recognition models. Inspired by the success of diffusion models in addressing visual domain sample imbalances, this paper introduces a new approach that utilizes the Markov Transfer Field (MTF) method for time series data visualization. This visualization, when combined with the Denoising Diffusion Probabilistic Model (DDPM), effectively enhances sample data and mitigates noise within the original dataset. Additionally, a transformer-based model tailored for time series visualization and air target intent recognition is developed. Comprehensive experimental results, encompassing comparative, ablation, and denoising validations, reveal that the proposed method achieves a notable 98.86% accuracy in air target intent recognition while demonstrating exceptional robustness and generalization capabilities. This approach represents a promising avenue for advancing air target intent recognition.

Text sentiment recognition model based on BERT and dynamic ensemble selection

Text sentiment recognition model based on BERT and dynamic ensemble selection

A Novel Lower-Limb Motion Intention Recognition Model Using ConvNet with Prob-sparse Attention

A Novel Lower-Limb Motion Intention Recognition Model Using ConvNet with Prob-sparse Attention

An EEG-based emotion recognition method by fusing multi-frequency-spatial features under multi-frequency bands.

Recognition of emotion changes is of great significance to a person's physical and mental health. At present, EEG-based emotion recognition methods are mainly focused on time or frequency domains, but rarely on spatial information. Therefore, the goal of this study is to improve the performance of emotion recognition by integrating frequency and spatial domain information under multi-frequency bands. Firstly, EEG signals of four frequency bands are extracted, and then three frequency-spatial features of differential entropy (DE) symmetric difference (SD) and symmetric quotient (SQ) are separately calculated. Secondly, according to the distribution of EEG electrodes, a series of brain maps are constructed by three frequency-spatial features for each frequency band. Thirdly, a Multi-Parallel-Input Convolutional Neural Network (MPICNN) uses the constructed brain maps to train and obtain the emotion recognition model. Finally, the subject-dependent experiments are conducted on DEAP and SEED-IV datasets. The experimental results of DEAP dataset show that the average accuracy of four-class emotion recognition, namely, high-valence high-arousal, high-valence low-arousal, low-valence high-arousal and low-valence low-arousal, reaches 98.71 %. The results of SEED-IV dataset show the average accuracy of four-class emotion recognition, namely, happy, sad, neutral and fear reaches 92.55 %. This method has a best classification performance compared with the state-of-the-art methods on both four-class emotion recognition datasets. This EEG-based emotion recognition method fused multi-frequency-spatial features under multi-frequency bands, and effectively improved the recognition performance compared with the existing methods.

Machine learning-based cyber attack recognition model

Machine learning-based cyber attack recognition model

A novel technique for rapid determination of pressure injury stages using intelligent machine vision

A developed intelligent machine vision system combined with deep-learning algorithms was attempted to determine pressure injury (PI) stages rapidly. A total of 500 images were selected according to the color and texture characteristics of probable PI sites closely related to fie PI stages based on the guidance of PI experts. Each target box of the PI site was labeled by the same researcher for label consistency. Characteristic values of pressure injuries were extracted from segmented images for further model construction. In developing the rapid determination models, five you just look once (YOLO) pattern recognition models (i.e., YOLO8n, YOLO8s, YOLO8m, YOLO8l, and YOLO8x) were constructed, and they were optimized among 100 epochs. Compared with other models, the YOLO8l model showed the best result, with the precision values among pressure injury stage I to V (i.e., PI_I, PI_II, PI_III, PI_IV, and PI_V) of 0.98, 0.97, 0.95, 0.95, and 0.94, respectively. The overall results suggest that this intelligent machine vision system is useful for PI stage determination and perhaps other disease diagnoses closely related to color and texture characteristics.

CT-RURnet: a novel network design for radar unmanned aerial vehicles recognition

Abstract The rapid growth of commercial unmanned aerial vehicles (UAVs) has led to more incidents. These include illegal activities by malicious actors. Such activities pose significant threats to public facilities and people’s safety. Therefore, distinguishing UAVs from other targets is a critical measure in preventing potential hazards. This paper proposes an enhanced feature extraction and global modeling model for UAV recognition, named CT-RURnet. The model directly learns from radar echo data, firstly using the convolutional neural networks module to quickly extract local features, and then using the Transformer module to establish global connections among these local features. The proposed model primarily focuses on learning the micro-Doppler features produced by targets, which facilitates UAV recognition. The used datasets include both simulated data and the Real Doppler RAD-DAR (RDRD) dataset. The proposed network achieves accuracy rates of 98.7% (10 SNR), 97.9% (5 SNR), and 86.95% (0 SNR) on the simulated dataset under varying SNR conditions and 97.14% on the real dataset. Compared to other baseline models, the CT-RURnet consistently delivers superior results. Ablation experiments are conducted to verify the contribution of each module to the overall network performance.

Ge’ez Digit Recognition Model Based on Convolutional Neural Network

ABSTRACT Despite the historical significance of the Ge’ez script, there is a notable scarcity of studies on Ge’ez digit recognition, compounded by the challenges posed by the absence of publicly accessible datasets. The complex structure of Ge’ez digits further complicates the recognition task. In response to this gap, our study addresses the development of a digit recognition model based on deep learning (DL) specifically tailored for printed Ge’ez digits, accompanied by the creation of a comprehensive study dataset. The proposed model architecture seamlessly integrates one input layer, six convolutional layers, and three Max-Pooling layers. To assess the model’s performance, we meticulously curated a Ge’ez digit dataset comprising 72,000 images, well-suited for DL applications using the ocropus-linegen model within OCRopus, a free document analysis tool. Leveraging cutting-edge DL algorithms, our proposed model demonstrates an impressive accuracy of 97.29%. This surpasses the performance of previous Ge’ez digit recognition models, marking a noteworthy advancement in this underexplored domain.

A continual learning framework to train robust image recognition models by adversarial training and knowledge distillation

Deep learning has been widely adopted in many image recognition tasks with great success. It has now been applied to conducting tasks on vision-based edge devices with resource limitation. To securely deploy image services on such devices, this work develops a framework to investigate the key issues relate the most to training robust and lightweight models. The primary issue is about safety. It is well known that the deep learning method is vulnerable to well-designed attacks. To meet the challenge of continuously evolving attacks, we develop a new defensive approach that integrates continual learning and adversarial training to improve the defensive model's corruption robustness and structure compactness. This approach adopts the structure of a progressive neural model to establish a robust model over time. The second issue is about how to train a lightweight model. Regarding this, we develop another new knowledge distillation approach that integrates both self-distillation and multi-teacher distillation techniques. With such a specially designed student-teacher learning structure and new loss functions, our approach is thus adaptable to different task situations. To verify the proposed framework, we have conducted a series of experiments to evaluate our approach. The results well confirm its functions and effectiveness.

基于轻量化改进YOLOv5s模型的稻瘟病检测方法及移动部署

For achieving more efficient recognition results and deployment on mobile devices, a rice blast recognition model was constructed by making lightweight improvements to YOLOv5s. First, using YOLOv5s as the base, GhostConv was introduced to replace standard convolution in its backbone and neck, and LightC3 module was built to improve the C3 module in the neck. This significantly reduced the computational burden and model size. Furthermore, Concat operator was replaced with BiFPN and SE attention mechanism was integrated to maintain accuracy when reducing model complexity. These modifications enhanced the model's ability to capture fine-grained features and multi-scale information. The experimental results showed that the designed model had a 49% decrease in the number of model parameters and a 50% decrease in FLOPs without a decrease in precision on self-built rice blast dataset, compared with the YOLOv5s, achieving the good balance between detection performance and model lightweight. Then, an APP named RiceBlastDetector was built based on the model, achieving accurate detection in the scenario with the different characterization scale disease spots from experiments in the field, which can provide a reference for detecting other crop diseases.

Literature Review of Research on English Abstracts for Scientific Papers

This paper reviews the research status and progress in the field of English abstracts of scientific papers, introduces the research background and importance of English abstracts of scientific papers as an important part of academic papers, and points out the key role of English abstracts of scientific papers in the improvement of academic communication and international influence. Through combing and analyzing the related literature, the main achievements and progresses in the research of English abstracts of scientific papers are elaborated in detail, including the analysis of the move structure, the characteristics of academic vocabulary usage, the construction of the lexicon of semantic features, and the comparison of lexical chunks, etc., and the current status and characteristics of the research in the field in China are also introduced. This paper finds that there are some hot and difficult issues in the research field of English abstracts of scientific papers, mainly in the areas of standardized writing of abstracts, cross-cultural comparison of abstract linguistic features, and the construction of automatic citation and recognition models. Finally, a comprehensive review of the existing research is conducted, pointing out the shortcomings of the current stage of research, such as the limited number of samples and the feature lexicon vocabulary to be expanded, etc., and proposing the directions and possible development trends of future research, such as the study of the writing process and cognitive mechanism, the deepening of interdisciplinary and cross-cultural contrastive research, and the application and innovation of the natural language processing technology. This paper aims to provide reference for researchers in the field of English abstracts for scientific papers, and to provide ideas and suggestions for follow-up research.

Functional anatomy of zinc finger antiviral protein complexes

AbstractZAP is an antiviral protein that binds to and depletes viral RNA, which is often distinguished from vertebrate host RNA by its elevated CpG content. Two ZAP cofactors, TRIM25 and KHNYN, have activities that are poorly understood. Here, we show that functional interactions between ZAP, TRIM25 and KHNYN involve multiple domains of each protein, and that the ability of TRIM25 to multimerize via its RING domain augments ZAP activity and specificity. We show that KHNYN is an active nuclease that acts in a partly redundant manner with its homolog N4BP1. The ZAP N-terminal RNA binding domain constitutes a minimal core that is essential for antiviral complex activity, and we present a crystal structure of this domain that reveals contacts with the functionally required KHNYN C-terminal domain. These contacts are remote from the ZAP CpG binding site and would not interfere with RNA binding. Based on our dissection of ZAP, TRIM25 and KHNYN functional anatomy, we could design artificial chimeric antiviral proteins that reconstitute the antiviral function of the intact authentic proteins, but in the absence of protein domains that are otherwise required for activity. Together, these results suggest a model for the RNA recognition and action of ZAP-containing antiviral protein complexes.

Aggressive Driving Behavior Recognition based on K-means Clustering

Accurate identification of aggressive driving behavior is crucial for improving traffic safety and promoting social responsibility, and it is also a key component in the realization of intelligent transportation systems. This paper proposes a method for aggressive driving behavior recognition and data modeling based on K-means clustering. First, the NGSIM (Next Generation Simulation) vehicle trajectory dataset is used for preprocessing the raw data, and effective behavioral feature representations are obtained through feature extraction. Next, Principal Component Analysis (PCA) is applied to reduce the dimensionality of the data, extracting three main components to simplify the data structure. Subsequently, the K-means clustering algorithm is employed to classify the reduced data and identify potential patterns of aggressive driving behavior. Experimental results demonstrate that the proposed method effectively identifies aggressive driving behavior, providing technical support for improving traffic system safety and raising awareness of social responsibility.