Convolutional Neural Network Features Research Articles

Scientific text citation analysis using CNN features and ensemble learning model.

Citation illustrates the link between citing and cited documents. Different aspects of achievements like the journal's impact factor, author's ranking, and peers' judgment are analyzed using citations. However, citations are given the same weight for determining these important metrics. However academics contend that not all citations can ever have equal weight. Predominantly, such rankings are based on quantitative measures and the qualitative aspect is completely ignored. For a fair evaluation, qualitative evaluation of citations is needed in addition to quantitative ones. Many existing works that use qualitative evaluation consider binary class and categorize citations as important or unimportant. This study considers multi-class tasks for citation sentiments on imbalanced data and presents a novel framework for sentiment analysis in in-text citations of research articles. In the proposed technique, features are retrieved using a convolutional neural network (CNN), and classification is performed using a voting classifier that combines Logistic Regression (LR) and Stochastic Gradient Descent (SGD). The class imbalance problem is handled by the synthetic minority oversampling technique (SMOTE). Extensive experiments are performed in comparison with the proposed approach using SMOTE-generated data and machine learning models by term frequency (TF), and term frequency-inverse document frequency (TF-IDF) to evaluate the efficacy of the proposed approach for citation analysis. It is found that the proposed voting classifier using CNN features achieves an accuracy, precision, recall, and F1 score of 0.99 for all. This work not only advances the field of sentiment analysis in academic citations but also underscores the importance of incorporating qualitative aspects in evaluating the impact and sentiments conveyed through citations.

Semi-supervised correction model for turbulence-distorted images

Significant progress has been made in addressing turbulence distortion in recent years, but persistent challenges remain. Firstly, existing methods heavily rely on fully supervised optimization strategies and synthetic datasets, posing difficulties in effectively utilizing unlabeled real data for training. Secondly, most approaches construct networks in a straightforward manner, overlooking the representation model of phase distortion and point spread function (PSF) in spatial and channel dimensions. This oversight restricts the potential for distortion correction. To address these challenges, this paper proposes a semi-supervised atmospheric turbulence correction method based on the mean-teacher framework. Our approach imposes constraints on the unlabeled data of student networks using pseudo-labels generated by teacher networks, thereby enhancing the generalization ability by leveraging information from unlabeled data. Furthermore, we introduce to use no-reference image quality assessment criterion to select the most reliable pseudo-label for each unlabeled sample by predicting physical parameters that indicating the level of degradation. Additionally, we propose to combine sliding window-based self-attention with channel attention to facilitate local-global context interaction. This design is inspired by the representation of phase distortion and PSF, which can be characterized by coefficients and basis functions corresponding to the channel-wise representation of convolutional neural network features. Moreover, the base functions exhibit spatial correlation, akin to Zenike and Airy disks. Experimental results show that the proposed method surpasses state-of-the-art models.

AI student success predictor: Enhancing personalized learning in campus management systems

Campus Management Systems (CMSs) are vital tools in managing educational institutions, handling tasks like student enrollment, scheduling, and resource allocation. The increasing adoption of CMS for online and mixed-learning environments highlights their importance. However, inherent limitations in conventional CMS platforms hinder personalized student guidance and effective identification of academic challenges. Addressing this crucial gap, our study introduces an AI Student Success Predictor empowered by advanced machine learning algorithms, capable of automating grading processes, predicting student risks, and forecasting retention or dropout outcomes. Central to our approach is the creation of a standardized dataset, meticulously curated by integrating student information from diverse relational databases. A Convolutional Neural Network (CNN) feature learning block is developed the extract the hidden patterns in the student data. This classification model stands as an ensemble masterpiece, incorporating SVM, Random Forest, and KNN classifiers, subsequently refined by a Bayesian averaging model. The proposed ensemble model shows the ability to predict the student grades, retention, and risk levels of dropout. The accuracy achieved by the proposed model is assessed using test data, culminating in a commendable 93% accuracy for student grade prediction and student risk prediction, and a solid 92% accuracy for the complex domain of retention and dropout forecasting. The proposed AI system seamlessly integrates with existing CMS infrastructure, enabling real-time data retrieval and swift, accurate predictions, enhancing academic decision-making efficiency. Our study's pioneering AI Student Success Predictor bridges the chasm between traditional CMS limitations and the growing demands of modern education.

Detection of Drowsy Drivers in Video Sequences via LSTM with CNN Features

Fatigued driving is one of the main causes of traffic accidents. This paper developed a method for sleep drive detection that makes use of the convolutional neural network (CNN), recurrent neural network (RNN), short-term neural network (LSTM), cascade, and cam switching algorithms. Using CNN, the algorithm initially extracts the driver's face traits. The RNN that predicts the driver's degree of tiredness is then fed these features. The technology tracks the driver's eyes and eyelids using algorithms called Cascades and Cam shifts. The accuracy of the sleepiness prediction is increased further with the usage of this data. Using datasets of video recordings of attentive and somnolent drivers, the system was assessed. The outcomes demonstrated how well the algorithm could identify sleepy drivers. This system may be able to stop sleepy driving-related accidents. Ten thousand videos of awake and sleepy drivers were used to train the algorithm. Drivers can be observed in real time by the system. If the system detects that the driver is sleepy, it has the ability to sound an alarm. The technology can be used to stop driving when sleepy from causing accidents. Keywords: Convolution neural network, Recurrent neural networks, long short-term memory, eye tracking, face detection, drowsy driver detection, deep learning, computer vision, machine learning, and deep learning.

EEG classification model for virtual reality motion sickness based on multi-scale CNN feature correlation

BackgroundVirtual reality motion sickness (VRMS) is a key issue hindering the development of virtual reality technology, and accurate detection of its occurrence is the first prerequisite for solving the issue. ObjectiveIn this paper, a convolutional neural network (CNN) EEG detection model based on multi-scale feature correlation is proposed for detecting VRMS. MethodsThe model uses multi-scale 1D convolutional layers to extract multi-scale temporal features from the multi-lead EEG data, and then calculates the feature correlations of the extracted multi-scale features among all the leads to form the feature adjacent matrixes, which converts the time-domain features to correlation-based brain network features, thus strengthen the feature representation. Finally, the correlation features of each layer are fused. The fused features are then fed into the channel attention module to filter the channels and classify them using a fully connected network. Finally, we recruit subjects to experience 6 different modes of virtual roller coaster scenes, and collect resting EEG data before and after the task to verify the model. Results: The results show that the accuracy, precision, recall and F1-score of this model for the recognition of VRMS are 98.66 %, 98.65 %, 98.68 %, and 98.66 %, respectively. The proposed model outperforms the current classic and advanced EEG recognition models. SignificanceIt shows that this model can be used for the recognition of VRMS based on the resting state EEG.

Convolutional neural networks for urban green areas semantic segmentation on Sentinel-2 data

Urban green areas are essential components of any urban environment, providing a wide range of uses. However, there is currently a noticeable absence of an automated tool for their land use classification. The use of urban green areas depends on their size, shape, and relationship with their surroundings, all of which are fundamental features in convolutional neural networks. Various convolutional neural network architectures (FCN, U-Net, SegNet, DeepLabv3+) were evaluated on open and widely accessible Sentinel-2 data for semantic segmentation of land cover and land use in different levels of urban green areas nomenclature and band combinations. Moreover, we compared the CNNs with random forests model as a baseline to underline the CNNs’ strengths. The evaluation found that convolutional neural networks are capable of the land cover and land use semantic segmentation not only on the full-band Sentinel-2 scenes but also on limited subdatasets consisting only of RGB bands. U-Net is identified as the best-performing architecture, achieving an overall accuracy of almost 95% for a simple binary vegetation detection, 90% for the land-use task, and almost 88% for the land-use task enhanced by a distinction between high and low vegetation, while random forests reached 93%, 84%, and 81%, respectively. CNNs’ misclassifications were primarily identified at the edges of two neighboring competing classes where mixed pixels appear. Data augmentation improved the model’s performance in 94% of cases. However, dropout layers led to an overall accuracy decrease in more than half of the cases. Additionally, we compared the segmented urban green area with a pan-European dataset, the Coordination of Information on the Environment Land Cover - and found that the latter omits 74% of the total urban vegetation. This is mainly due to its minimal mapping unit specification. It is concluded that the most suitable approach for automated urban green areas land cover and land use semantic segmentation is the use of convolutional neural networks, from the tested architectures particularly U-Net.

Multi‐object road waste detection and classification based on binocular vision

AbstractA road multi‐object detection algorithm is one of the core algorithms for intelligent road cleaning robots relying on machine vision. Most existing object detection algorithms analyse all image regions and finally calculate the category and location of each object. However, it is not necessary to analyse all areas of the image when detecting objects on the road surface where the background changes little, and the number of objects is small because there will be a lot of invalid calculations. If we can perform targeted local analysis on images instead of analysing all image regions, it will improve the detection efficiency. Therefore, this paper proposes a multi‐object detection method using a binocular camera and a convolutional neural network (CNN) that effectively reduces invalid calculations during the detection and improves detection efficiency. In the developed method, the binocular vision image acquired by the binocular camera is stereo matched and equalized, while linear regression and coordinate transformation eliminate the angle of the camera pair concerning the road surface. Then, the coordinates of the regions of interest (ROI) is calculated in the left vision image and the features within the ROI is extracted from the corresponding CNN's feature map. Next, ROI pooling resizes the extracted feature maps of different sizes to the same size, which are then input to the fully connected layers to output the results. The proposed binocular network and faster R‐CNN (VGG16) are trained and tested on a dataset involving 1000 road waste images. The experimental results demonstrate that the developed binocular network improves the detection accuracy and speed by 28.56% and 78.39%, respectively, compared with faster R‐CNN (VGG16), providing a reliable basis for a machine vision‐based intelligent road cleaning robot.

A hybrid 1D CNN-BiLSTM model for epileptic seizure detection using multichannel EEG feature fusion

Epilepsy, a chronic non-communicable disease is characterized by repeated unprovoked seizures, which are transient episodes of abnormal electrical activity in the brain. While Electroencephalography (EEG) is considered as the gold standard for diagnosis in current clinical practice, manual inspection of EEG is time consuming and biased. This paper presents a novel hybrid 1D CNN-Bi LSTM feature fusion model for automatically detecting seizures. The proposed model leverages spatial features extracted by one dimensional convolutional neural network and temporal features extracted by bi directional long short-term memory network. Ictal and inter ictal data is first acquired from the long multichannel EEG record. The acquired data is segmented and labelled using small fixed windows. Signal features are then extracted from the segments concurrently by the parallel combination of CNN and Bi-LSTM. The spatial and temporal features thus captured are then fused to enhance classification accuracy of model. The approach is validated using benchmark CHB-MIT dataset and 5-fold cross validation which resulted in an average accuracy of 95.90%, with precision 94.78%, F1 score 95.95%. Notably model achieved average sensitivity of 97.18% with false positivity rate at 0.05/hr. The significantly lower false positivity and false negativity rates indicate that the proposed model is a promising tool for detecting seizures in epilepsy patients. The employed parallel path network benefits from memory function of Bi-LSTM and strong feature extraction capabilities of CNN. Moreover, eliminating the need for any domain transformation or additional preprocessing steps, model effectively reduces complexity and enhances efficiency, making it suitable for use by clinicians during the epilepsy diagnostic process.

Prediction and Classifications of Breast Cancer using Enhanced Convolutional Neural Network Approaches

In worldwide women mortality increases extremely every year due to breast cancer and diagnosis of the issue through prediction is very much imperative for healthy lifespan. Here precision of cancer extrapolation is an essential thing for survivability of patient with appropriate treatment. Deep learning algorithms have materialised as influential tool for predicting breast cancer in medical image processing, which leverages capabilities of artificial neural networks (ANN) that are intended to mimic an architecture and functionalities of human brain. Superior features of convolutional neural network (CNN) in deep learning for handling image-based data like, exploiting spatial information, hierarchical feature learning, parameter sharing and data augmentation are important parameters in medical image processing. In this paper CNN algorithm is incorporated for predicting breast cancer in earlier and malignant stage, the results are compared with other deep learning algorithms and our proposed algorithm is expected to give better performance in parameters like accuracy testing, image classifiers, gene sequence classifiers and malignancy detection.

Vehicle-Type Recognition Method for Images Based on Improved Faster R-CNN Model.

The rapid increase in the number of vehicles has led to increasing traffic congestion, traffic accidents, and motor vehicle crime rates. The management of various parking lots has also become increasingly challenging. Vehicle-type recognition technology can reduce the workload of humans in vehicle management operations. Therefore, the application of image technology for vehicle-type recognition is of great significance for integrated traffic management. In this paper, an improved faster region with convolutional neural network features (Faster R-CNN) model was proposed for vehicle-type recognition. Firstly, the output features of different convolution layers were combined to improve the recognition accuracy. Then, the average precision (AP) of the recognition model was improved through the contextual features of the original image and the object bounding box optimization strategy. Finally, the comparison experiment used the vehicle image dataset of three vehicle types, including cars, sports utility vehicles (SUVs), and vans. The experimental results show that the improved recognition model can effectively identify vehicle types in the images. The AP of the three vehicle types is 83.2%, 79.2%, and 78.4%, respectively, and the mean average precision (mAP) is 1.7% higher than that of the traditional Faster R-CNN model.

Improved Convolutional Neural Network–Time-Delay Neural Network Structure with Repeated Feature Fusions for Speaker Verification

The development of deep learning greatly promotes the progress of speaker verification (SV). Studies show that both convolutional neural networks (CNNs) and dilated time-delay neural networks (TDNNs) achieve advanced performance in text-independent SV, due to their ability to sufficiently extract the local feature and the temporal contextual information, respectively. Also, the combination of the above two has achieved better results. However, we found a serious gridding effect when we apply the 1D-Res2Net-based dilated TDNN proposed in ECAPA-TDNN for SV, which indicates discontinuity and local information losses of frame-level features. To achieve high-resolution process for speaker embedding, we improve the CNN–TDNN structure with proposed repeated multi-scale feature fusions. Through the proposed structure, we can effectively improve the channel utilization of TDNN and achieve higher performance under the same TDNN channel. And, unlike previous studies that have all converted CNN features to TDNN features directly, we also studied the latent space transformation between CNN and TDNN to achieve efficient conversion. Our best method obtains 0.72 EER and 0.0672 MinDCF on VoxCeleb-O test set, and the proposed method performs better in cross-domain SV without additional parameters and computational complexity.

Super-resolution of diffusion-weighted images using space-customized learning model.

Diffusion-weighted imaging (DWI) is a noninvasive method used for investigating the microstructural properties of the brain. However, a tradeoff exists between resolution and scanning time in clinical practice. Super-resolution has been employed to enhance spatial resolution in natural images, but its application on high-dimensional and non-Euclidean DWI remains challenging. This study aimed to develop an end-to-end deep learning network for enhancing the spatial resolution of DWI through post-processing. We proposed a space-customized deep learning approach that leveraged convolutional neural networks (CNNs) for the grid structural domain (x-space) and graph CNNs (GCNNs) for the diffusion gradient domain (q-space). Moreover, we represented the output of CNN as a graph using correlations defined by a Gaussian kernel in q-space to bridge the gap between CNN and GCNN feature formats. Our model was evaluated on the Human Connectome Project, demonstrating the effective improvement of DWI quality using our proposed method. Extended experiments also highlighted its advantages in downstream tasks. The hybrid convolutional neural network exhibited distinct advantages in enhancing the spatial resolution of DWI scans for the feature learning of heterogeneous spatial data.

Anomaly Detection of Permanent Magnet Synchronous Motor Based on Improved DWT-CNN Multi-Current Fusion.

The Permanent Magnet Synchronous Motor (PMSM) is the power source maintaining the stable and efficient operation of various pieces of equipment; hence, its reliability is crucial to the safety of public equipment. Convolutional Neural Network (CNN) models face challenges in extracting features from PMSM current data. A new Discrete Wavelet Transform Convolutional Neural Networks (DW-CNN) feature with fusion weight updating Long Short-Term Memory (LSTM) anomaly detection is proposed in this paper. This approach combines Discrete Wavelet Transform (DWT) with high and low-frequency separation processing and LSTM. The anomaly detection method adopts DWT and CNN by separating high and low-frequency processing. Moreover, this method combines the hybrid attention mechanism to extract the multi-current signal features and detects anomalies based on weight updating the LSTM network. Experiments on the motor bearing real fault dataset and the PMSM stator fault dataset prove the method's strong capability in fusing current features and detecting anomalies.

Vision transformers are active learners for image copy detection

Image Copy Detection (ICD) is developed to identify and track duplicated or manipulated images. The majority of existing methods rely on Convolutional Neural Networks (CNNs) and are trained using unsupervised learning techniques, which leads to subpar performance. We discover that by carefully designing the training process, Vision Transformer (ViT) backbones yield superior results. Specifically, directly training a ViT for ICD often leads to overfitting on the training images, which in turn results in poor generalization to unseen (test) images. Consequently, we initially train a CNN (such as ResNet-50), and during the ViT training, the distances between the features of CNN and ViT are regularized. We also incorporate an active learning method to further enhance performance. Notably, due to the visual discrepancy between auto-generated transformations and those used in the query set, we incorporate a small number (approximately 0.5% of unlabeled training images) of manually produced and labeled positive pairs. Training models on these pairs results in a significant performance boost though with little cost. Experimental findings demonstrate the effectiveness of our approach, and our method achieves state-of-the-art performance. Our code is available at: https://github.com/WangWenhao0716/ViT4ICD.

Optimised ResNet50 for Multi-Class Classification of Brain Tumors

Categorizing brain cancers, including glioma, meningioma, and pituitary tumors, based on magnetic resonance imaging (MRI) images presents a significant challenge. Deep learning and machine learning techniques have shown promise in enhancing image categorization. To address this challenge, we leverage the power of the optimized ResNet50 model. Our approach involves classifying medical images using Convolutional Neural Network (CNN) features, which are then compared with the ResNet50 model. The primary goal is to detect brain tumors at an early stage using an advanced deep-learning model. We utilize an accessible dataset from Figshare, containing MRI images of the three distinct categories of brain tumors. Existing brain tumor models face limitations in handling multi-class problems and early-stage diagnosis. Therefore, we propose a fully automated approach employing Convolutional Neural Networks (CNN) to extract diverse properties from brain MRI scans. This method aims to provide accurate tumor diagnosis, even with a high number of classes and limited information in MRI data. Our proposed model involves the creation of identification blocks within a four-layered primary architecture, followed by testing and assessment of the interconnected layers. The results demonstrate that our model outperforms existing methods, achieving an impressive overall classification accuracy of 99.03%,

CRIECNN: Ensemble convolutional neural network and advanced feature extraction methods for the precise forecasting of circRNA-RBP binding sites

Circular RNAs (circRNAs) have surfaced as important non-coding RNA molecules in biology. Understanding interactions between circRNAs and RNA-binding proteins (RBPs) is crucial in circRNA research. Existing prediction models suffer from limited availability and accuracy, necessitating advanced approaches. In this study, we propose CRIECNN (Circular RNA-RBP Interaction predictor using an Ensemble Convolutional Neural Network), a novel ensemble deep learning model that enhances circRNA-RBP binding site prediction accuracy. CRIECNN employs advanced feature extraction methods and evaluates four distinct sequence datasets and encoding techniques (BERT, Doc2Vec, KNF, EIIP). The model consists of an ensemble convolutional neural network, a BiLSTM, and a self-attention mechanism for feature refinement. Our results demonstrate that CRIECNN outperforms state-of-the-art methods in accuracy and performance, effectively predicting circRNA-RBP interactions from both full-length sequences and fragments. This novel strategy makes an enormous advancement in the prediction of circRNA-RBP interactions, improving our understanding of circRNAs and their regulatory roles.

Estimation of aquatic ecosystem health using deep neural network with nonlinear data mapping

Estimation of aquatic ecosystem health indices can assist in reducing the burden of time-consuming, labor-intensive, and cost-effective fieldwork for the sustainable evaluation of freshwater ecosystem status. In this study, we developed a deep neural network to estimate the trophic diatom index (TDI), benthic macroinvertebrate index (BMI), and fish assessment index (FAI) using water quality and hydraulic and hydrological data. A convolutional neural network (CNN) model was built to estimate health indices. In addition, an autoencoder was adopted to produce manifold features that were used as inputs for the CNN model. Conventional machine learning models, including artificial neural networks, support vector machines, random forests, and extreme gradient boosting, have been developed to estimate the TDI, BMI, and FAI. The results showed that the CNN with an autoencoder exhibited the best performance, with validation accuracies of Nash Sutcliffe Efficiency (NSE) and root mean squared error (RMSE) values of 0.592 and 17.249 for TDI, 0.669 and 12.282 for BMI, and 0.638 and 13.897 for FAI, respectively. The autoencoder enhanced the nonlinear feature learning of the time series and static input data, which contributed to improving the CNN feature extraction for accurate estimation of aquatic ecosystem health indices compared to other data-driven approaches. Therefore, deep learning techniques can be used to investigate aquatic ecosystem health by successfully reflecting the quantitative and qualitative features of health indices.

A state estimation method based on CNN-LSTM for ball screw

Ball screw is widely used in the engineering field, and accurate estimation of their state is crucial for the reliability of system operation. However, existing methods often overlook the time series characteristics and spatial correlation of vibration signals, unable to provide complete degradation information and divide the degradation process, resulting in limited prediction accuracy. Therefore, a state estimation method for ball screw based on Convolutional Neural Networks (CNN) and Long Short-Term Memory Neural Networks (LSTM) is proposed. An experiment of ball screw transmission equipment was conducted to collect vibration signals throughout the entire life cycle and verify the proposed method. Firstly, the frequency domain amplitude signal of the transformed ball screw is normalized to eliminate scale differences, which serves as the input for CNN feature extraction. Then, these deep features are input into the LSTM network to capture the fault evolution patterns that reveal the degradation of ball screw performance, and achieve accurate estimation of ball screw state. The final prediction accuracy was 97.87%, verifying the effectiveness of the proposed method.

Fine-grained textural detail enhancement: concatenating convolutional neural network features with adaptive fuzzy logic

Fine-grained textural detail enhancement: concatenating convolutional neural network features with adaptive fuzzy logic

Combinatorial CNN-Transformer Learning with Manifold Constraints for Semi-supervised Medical Image Segmentation

Semi-supervised learning (SSL), as one of the dominant methods, aims at leveraging the unlabeled data to deal with the annotation dilemma of supervised learning, which has attracted much attentions in the medical image segmentation. Most of the existing approaches leverage a unitary network by convolutional neural networks (CNNs) with compulsory consistency of the predictions through small perturbations applied to inputs or models. The penalties of such a learning paradigm are that (1) CNN-based models place severe limitations on global learning; (2) rich and diverse class-level distributions are inhibited. In this paper, we present a novel CNN-Transformer learning framework in the manifold space for semi-supervised medical image segmentation. First, at intra-student level, we propose a novel class-wise consistency loss to facilitate the learning of both discriminative and compact target feature representations. Then, at inter-student level, we align the CNN and Transformer features using a prototype-based optimal transport method. Extensive experiments show that our method outperforms previous state-of-the-art methods on three public medical image segmentation benchmarks.