Combination Of Convolutional Neural Network Research Articles

HD-Former: A hierarchical dependency Transformer for medical image segmentation

Medical image segmentation is a compelling fundamental problem and an important auxiliary tool for clinical applications. Recently, the Transformer model has emerged as a valuable tool for addressing the limitations of convolutional neural networks by effectively capturing global relationships and numerous hybrid architectures combining convolutional neural networks (CNNs) and Transformer have been devised to enhance segmentation performance. However, they suffer from multilevel semantic feature gaps and fail to account for multilevel dependencies between space and channel. In this paper, we propose a hierarchical dependency Transformer for medical image segmentation, named HD-Former. First, we utilize a Compressed Bottleneck (CB) module to enrich shallow features and localize the target region. We then introduce the Dual Cross Attention Transformer (DCAT) module to fuse multilevel features and bridge the feature gap. In addition, we design the broad exploration network (BEN) that cascades convolution and self-attention from different percepts to capture hierarchical dense contextual semantic features locally and globally. Finally, we exploit uncertain multitask edge loss to adaptively map predictions to a consistent feature space, which can optimize segmentation edges. The extensive experiments conducted on medical image segmentation from ISIC, LiTS, Kvasir-SEG, and CVC-ClinicDB datasets demonstrate that our HD-Former surpasses the state-of-the-art methods in terms of both subjective visual performance and objective evaluation. Code: https://github.com/barcelonacontrol/HD-Former.

ENSEMBLE CNN WITH ADASYN FOR MULTICLASS CLASSIFICATION ON CABBAGE PESTS

Image classification is a complex process influenced by various factors, one of which is the amount of image data. In the context of cabbage pest classification, data often exhibits a significant class imbalance, where certain pests are more prevalent than others. This imbalance can pose challenges during model training and evaluation, potentially leading to biases in favor of the majority pests and reduced accuracy in identifying and classifying the less common ones. This research aims to enhance the classification performance for multiclass data specific to cabbage pests. We propose an ensemble learning approach that combines Convolutional Neural Network (CNN), Support Vector Machine (SVM), and Bagging methods. To address the imbalance issue inherent in cabbage pest data, we employ the Adaptive Synthetic Sampling (ADASYN) resampling technique. The CNN acts as the primary image identifier and classifier for various cabbage pests. Subsequently, the CNN model is integrated into SVM and Bagging models to mitigate the challenges of imbalanced data in pest classification. The research outcomes demonstrate that our ensemble approach, in conjunction with the ADASYN resampling technique, achieves an impressive accuracy rate of 97%, signifying its potential for improved cabbage pest detection and classification.

Improving brain tumor classification with combined convolutional neural networks and transfer learning

Brain tumors pose a serious threat, causing the deaths of thousands of people worldwide, and can lead to life-threatening consequences when not accurately diagnosed. The classification of brain tumors and the determination of correct treatment strategies are crucial, yet this process encounters various challenges. These tumors originate from different cell types and exhibit diversity in growth rates, histological features, and genetic structures. Some may show similarities at the microscopic level, complicating classification and making diagnosis and treatment challenging. The examination of brain MR images is a widely used method in diagnosing brain tumors. However, occasional misdiagnosis of tumors can lead to ineffective responses to treatments and reduced chances of survival for patients. Traditional machine learning classifiers require manually determined features, which is quite time-consuming. On the other hand, deep learning is highly effective in feature extraction and has recently been widely preferred in classification. In this context, the effectiveness of transfer learning architectures in brain tumor diagnosis was evaluated. Six different transfer learning architectures, including ResNet-50, MobileNet, VGG16, Inception-V3, DenseNet-121, and EfficientNetV2-M, were used in this study. A public MRI dataset was used for model validation and comparison with similar studies. To address the imbalance in the number of images among classes in the dataset, data augmentation techniques such as random rotation were applied during data preprocessing. Experiments revealed that the EfficientNetV2-M model outperformed other architectures with a 98.01% accuracy rate. Additionally, the study aimed to create a new model with more comprehensive feature extraction and generalization ability by combining the advantages of multiple models in different combinations. In this context, combinations of EfficientNetV2-M architecture with Inception-V3 and DenseNet-121 architectures were formed. Through these combinations, a concatenation-based EfficientNetV2-M + Inception-V3 model achieved a 98.41% accuracy value. It was observed that the proposed concatenation-based model outperformed advanced methods in improving medical imaging techniques and patient outcomes.

A Hybrid CNN-TransXNet Approach for Advanced Glomerular Segmentation in Renal Histology Imaging

In the specialized field of renal histology, precise segmentation of glomeruli in microscopic images is crucial for accurate clinical diagnosis and pathological analysis. Facing the challenge of discerning complex visual features, such as shape, texture, and size within these images, we introduce a novel segmentation model that innovatively combines convolutional neural networks (CNNs) with the advanced TransXNet block, specifically tailored for glomerular segmentation. This innovative model is designed to capture the intricate details and broader contextual features within the images, ensuring a comprehensive and precise segmentation process. The model's architecture unfolds in two primary phases: the down-sampling phase, which utilizes CNNs structures within the TransXNet block for meticulous extraction of detailed features, and the up-sampling phase, which employs CNNs deconvolution techniques to restore spatial resolution and enhance macroscopic feature representation. A critical innovation in our model is the implementation of residual connections between these two phases, which facilitate the seamless integration of features and minimize loss of precision during image reconstruction. Experimental results demonstrate a significant improvement in our model’s performance compared to existing medical image segmentation methods. We report enhancements in mean Pixel Accuracy (mPA) and mean Intersection over Union (mIoU), with increases of approximately 3–5% and 3–8%, respectively. Additionally, the segmented outputs exhibit higher subjective visual quality with fewer noise artifacts. These findings suggest that our model offers promising applications in the segmentation of medical microscopic images, marking a significant contribution to the domain.

Squeeze-and-excitation-attention-based mobile vision transformer for grading recognition of bladder prolapse in pelvic MRI images.

Bladder prolapse is a common clinical disorder of pelvic floor dysfunction in women, and early diagnosis and treatment can help them recover. Pelvic magnetic resonance imaging (MRI) is one of the most important methods used by physicians to diagnose bladder prolapse; however, it is highly subjective and largely dependent on the clinical experience of physicians. The application of computer-aided diagnostic techniques to achieve a graded diagnosis of bladder prolapse can help improve its accuracy and shorten the learning curve. The purpose of this study is to combine convolutional neural network (CNN) and vision transformer (ViT) for grading bladder prolapse in place of traditional neural networks, and to incorporate attention mechanisms into mobile vision transformer (MobileViT) for assisting in the grading of bladder prolapse. This study focuses on the grading of bladder prolapse in pelvic organs using a combination of a CNN and a ViT. First, this study used MobileNetV2 to extract the local features of the images. Next, a ViT was used to extract the global features by modeling the non-local dependencies at a distance. Finally, a channel attention module (i.e., squeeze-and-excitation network) was used to improve the feature extraction network and enhance its feature representation capability. The final grading of the degree of bladder prolapse was thus achieved. Using pelvic MRI images provided by a Huzhou Maternal and Child Health Care Hospital, this study used the proposed method to grade patients with bladder prolapse. The accuracy, Kappa value, sensitivity, specificity, precision, and area under the curve of our method were 86.34%, 78.27%, 83.75%, 95.43%, 85.70%, and 95.05%, respectively. In comparison with other CNN models, the proposed method performed better. Thus, the model based on attention mechanisms exhibits better classification performance than existing methods for grading bladder prolapse in pelvic organs, and it can effectively assist physicians in achieving a more accurate bladder prolapse diagnosis.

Data-driven fault detection framework for offshore wind-hydrogen systems

Offshore wind-hydrogen systems operate in harsh marine environments for extended periods, posing risks of low accessibility and high failure rates. This paper proposes a data-driven fault detection framework for offshore wind-hydrogen systems, aiming to promptly detect potential faults based on the system's physical parameters. The research establishes a comprehensive model based on the operating principles of offshore wind-hydrogen systems. The study uses fault injection techniques to simulate common partial faults in offshore wind-hydrogen systems and collects relevant physical parameters (such as wind turbine power generation, energy storage battery charging/discharging current, and electrolyzer hydrogen production rate). A fault detection model was constructed using a combination of Convolutional Neural Networks (CNN) and Bidirectional Long Short-Term Memory (BiLSTM) to enable precise detection of potential faults in the system. The established model provides theoretical support for designing, optimizing, and operating offshore wind-hydrogen systems, reducing reliance on actual experiments and lowering project risks and costs. Introducing artificial faults can evaluate the system's performance in different fault scenarios. This paper proposes a CNN-BiLSTM fault detection method that achieves automatic feature extraction and high-precision classification of time-series fault data, confirming the practicality and feasibility of offshore wind-hydrogen systems. This investigation realizes the integration of the offshore wind-hydrogen system model and deep learning, as well as a fully automated learning process from data to fault detection. By calculating and comparing results, it is verified that the framework has a high accuracy of 96.9%. This innovative research offers new methods and perspectives for enhancing the reliability and efficiency of fault detection in offshore wind-hydrogen systems.

Error-feedback three-phase optimization to configurable convolutional echo state network for time series forecasting

Time series forecasting is critical for many real-world applications. Convolutional echo state networks (CESNs) have shown intriguing time series modeling efficacy by combining convolutional neural network (CNN) and echo state network (ESN). However, current CESN models are tailored for the classification tasks and rely on elaborately designed neural architectures. To this end, we propose a novel configurable convolutional echo state network (CCESN) with an innovative error-feedback three-phase optimization (ETO) strategy for time series forecasting. The network is progressively constructed with heterogeneous modular subnetworks, including ESN, CNN, CESN, and reversed CESN modules. This scheme leverages the complementary feature extraction capabilities of convolutional and recurrent neural architectures. To adaptively evolve the CCESN, we propose a novel error-feedback three-phase optimization (ETO) strategy by selecting optimal subnetwork modules while step-wise tuning parameters. Comprehensive experiments are conducted on representative simulated and real-world datasets. The results indicate that ETO-CCESN can adaptively select and evolve heterogeneous subnetworks to acclimatize to varied scenarios, and thus demonstrate significant performance improvements, achieving a 45.69% average enhancement in forecasting accuracy compared to the existing CESN model, and surpassing the best baseline by 8.79% in terms of symmetric mean absolute percentage error across diverse forecasting tasks.

A federated learning-based zero trust intrusion detection system for Internet of Things

The rapid expansion of Internet of Things (IoT) devices presents unique challenges in ensuring the security and privacy of interconnected systems. As cyberattacks become more frequent, developing an effective and scalable Intrusion Detection System (IDS) based on Federated Learning (FL) for IoT becomes increasingly complex. Current methodologies struggle to balance spatial and temporal feature extraction, especially when dealing with dynamic and evolving cyber threats. The lack of diversity in datasets used for FL-based IDS evaluations further impedes progress. There is also a noticeable tradeoff between performance and scalability, particularly as the number of edge devices in communication increases. To address these challenges, this article introduces a horizontal FL model that combines Convolutional Neural Networks (CNN) and Bidirectional Long-Term Short Memory (BiLSTM) for effective intrusion detection. This hybrid approach aims to overcome the limitations of existing methods and enhance the effectiveness of intrusion detection in the context of FL for IoT. Specifically, CNN is used for spatial feature extraction, enabling the model to identify local patterns indicative of potential intrusions, while the BiLSTM component captures temporal dependencies and learns sequential patterns within the data. The proposed IDS follows a zero-trust model by keeping the data on local edge devices and sharing only the learned weights with the centralized FL server. The FL server then aggregates updates from various sources to optimize the accuracy of the global learning model. Experimental results using CICIDS2017 and Edge-IIoTset demonstrate the effectiveness of the proposed approach over centralized and federated deep learning-based IDS.

Fault Diagnosis of Gearbox Based on CNN and LSTM with Attention Mechanism

Gearbox is a key component of mechanical equipment, which has a complex structure, harsh working conditions, and a higher probability of failure. Therefore, gearbox fault diagnosis is vital to ensure the efficient operation of the whole mechanical equipment. However, traditional gearbox fault diagnosis methods mainly rely on manual feature extraction. To address this issue, a novel end-to-end fault diagnosis model that combines convolutional neural network (CNN), long short-term memory network (LSTM) and attention mechanism (AM) is proposed. Firstly, Spatial features are extracted from the original input data using CNNs, and then temporal features are extracted even further from the spatial features using LSTMs. The attention mechanism improves the network's attention to the global key features by assigning weights, and finally, the fault diagnosis results are derived from the fully connected layer and Softmax classifier. The experimental results show that the method is able to adaptively extract features and ensure higher diagnostic accuracy, validating the effectiveness of the proposed method.

Analysis of the influencing factors of students' vocational ability based on joint improvement of convolutional neural network and multimodal data

Students' professional ability includes innovation and entrepreneurship ability, independent ability, quality ability, professional ability and adaptability. The school's practical training teaching and professional setting, as well as the unique talent training model have a profound impact on training and development of students' professional ability. The continuous progress and development of society indicates that the competitiveness of occupational positions in the future will continue to increase. The talent market will also put forward newer and more difficult requirements for the quality of students. In study and life, cultivating students' good professional ability is the fundamental purpose of colleges and universities. Employers tend to pay more attention to overall quality of candidates when selecting employees. Whether students can enter the society smoothly after graduation has become an important task. It can be seen that it is very important to analyze the cultivation of students' professional ability and the influencing factors of professional ability. This work combines convolutional neural network (CNN) with the analysis of factors influencing students' vocational ability. First, this work proposes an improved residual block (IRB). It increases the width of the residual network by adding a residual connection channel to learn richer features. Second, this work proposes an improved attention module (IAM). It combines channel attention with spatial attention to improve model performance. Third, this work combines IRB and IAM to propose an SVAIFANet. The model can be applied to the analysis of factors influencing students' vocational ability. Fourth, this work conducts various experiments for proposed method, experimental data verify correctness for this method.

Paediatric Activity Anomaly Detection using CNN-XGBoost for Early Intervention and Patient Safety

The development of a robust anomaly detection system capable of distinguishing between normal and abnormal activities in paediatric care settings is of paramount importance for early intervention and patient safety. This endeavor presents substantial challenges due to the unique characteristics of paediatric activities, making it necessary to employ a combination of Convolutional Neural Networks (CNNs) and XGBoost to effectively address these complexities. The initial step involves the assembly of a meticulously annotated dataset comprising video sequences encompassing a broad spectrum of paediatric activities, both normal and abnormal, with a strong emphasis on diversity and representativeness. This dataset is subjected to rigorous preprocessing to ensure consistency and quality. This research employs a Kalman filter as a pre-processing step. This filter helps to reduce noise and enhance the quality of the video data. By smoothing and stabilizing the trajectories of objects or subjects within the video frames, the Kalman filter provides a cleaner input for subsequent stages of the process. CNN employed for feature extraction from video frames, capturing both spatial and temporal cues. To account for the temporal dimension inherent in video data, by ensuring that the model can capture nuanced patterns in paediatric activities. The accuracy of the proposed method is 88%. The extracted features or sequences are input into an XGBoost classifier, designed as a binary classification task to differentiate between normal and abnormal activities. Robust model evaluation is conducted on validation and test datasets using appropriate performance metrics, with continuous feedback loops established with healthcare professionals and caregivers to fine-tune and optimize the model's performance.

MAHyNet: Parallel Hybrid Network for RNA-Protein Binding Sites Prediction Based on Multi-Head Attention and Expectation Pooling.

RNA-binding proteins (RBPs) can regulate biological functions by interacting with specific RNAs, and play an important role in many life activities. Therefore, the rapid identification of RNA-protein binding sites is crucial for functional annotation and site-directed mutagenesis. In this work, a new parallel network that integrates the multi-head attention mechanism and the expectation pooling is proposed, named MAHyNet. The left-branch network of MAHyNet hybrids convolutional neural networks (CNNs) and gated recurrent neural network (GRU) to extract the features of one-hot. The right-branch network is a two-layer CNN network to analyze physicochemical properties of RNA base. Specifically, the multi-head attention mechanism is a computational collection of multiple independent layers of attention, which can extract feature information from multiple dimensions. The expectation pooling combines probabilistic thinking with global pooling. This approach helps to reduce model parameters and enhance the model performance. The combination of CNN and GRU enables further extraction of high-level features in sequences. In addition, the study shows that appropriate hyperparameters have a positive impact on the model performance. Physicochemical properties can be used to supplement characterization information to improving model performance. The experimental results show that MAHyNet has better performance than other models.

A hybrid approach using CNN and active contour model for automated segmentation of macular edema

Macula is the part of retina responsible for sharp and clear vision. Macular edema is caused by the accumulation of intraretinal fluid (IRF) in the macula, which is further distinguished by the compromised integrity of the blood-retinal barrier, particularly evident in the retinal vasculature. This results in swelling, that may lead to vision impairment and is the dominant sign of several ocular diseases, including age-related macular degeneration, diabetic retinopathy, etc. Quantitative analysis of the fluid regions in macular edema helps in ascertaining the severity as well as the response to treatment of the diseases. Optical coherence tomography (OCT) is a major tool used by ophthalmologists for visualizing edema. The prevalent practice for diagnosing and treating macular edema involves measuring Central Retinal Thickness (CRT). Segmenting the IRF in OCT images offers the potential for a more accurate and better quantification of macular edema. This paper proposes a novel method combining convolutional neural network (CNN) and active contour model for segmenting the IRF to ascertain the severity of macular edema. The IRF region is initially segmented using an encoder-decoder architecture. Contour evolution is then performed on this segmented image to demarcate the IRF boundaries. The advantage of the method is that it does not require precisely labeled images for training the CNN. A comparison of the experimental results with models employing CNN alone and with other state-of-the art methods demonstrates the superior performance and consistency of the proposed method.

Discerning Deception: A Face-Centric Deepfake Detection Approach with ResNeXt-50 and LSTMs

Abstract: A.I. has grown to epidemic proportions over the last years as its applied in almost all sectors to allocate workload from humans but end up being done effectively with no human intervention. A branch of A.I. called deep learning, which operates by mimicking human judgment and action through neural network systems. Nonetheless, with the increase height of the two platforms have been experienced sufficient cases of misguided individuals using tools to recycle videos, audios, and texts to achieve their agendas. This insinuates a due assumption that Generative Adversarial Networks, GANs, are central to the development of believable deepfakes. GANs have developed a crucial ability to generate videos that replace frames with material from another video source to create deepfakes videos. While GANs serves various purposes such as entertainment, teaching, and experimentation, malicious actors can misuse these deep learning techniques to manipulate videos, impacting the privacy of individuals in society. This paper conducts an analysis of different deepfake detection models, comparing their efficacy and discussing potential future extensions of deepfake technology. The study presents a novel deepfake detection approach utilizing a combination of Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN). This method utilizes ResNext50 for extracting features at the frame level, while employing LSTM (Long Short-Term Memory) for video classification based on these extracted features. Various datasets are incorporated, including the deepfake detection challenge dataset (DFDC) and Face Forensics deepfake collections (FF++), combining them to achieve a high-accuracy model capable of accurately discerning between real and deepfake videos. The results of this study make a valuable contribution to the continuous endeavors aimed at improving deepfake detection abilities and ensuring privacy protection in a time heavily influenced by artificial intelligence

Convolutional Neural Networks to Facilitate the Continuous Recognition of Arabic Speech with Independent Speakers

Automatic speech recognition (ASR) is a field of research that focuses on the ability of computers to process and interpret speech feedback from humans and to provide the highest degree of accuracy in recognition. Speech is one of the simplest ways to convey a message in a basic context, and ASR refers to the ability of machines to process and accept speech data from humans with the greatest degree of accuracy. As the human-to-machine interface continues to evolve, speech recognition is expected to become increasingly important. However, the Arabic language has distinct features that set it apart from other languages, such as the dialect and the pronunciation of words. Until now, insufficient attention has been devoted to continuous Arabic speech recognition research for independent speakers with a limited database. This research proposed two techniques for the recognition of Arabic speech. The first uses a combination of convolutional neural network (CNN) and long short-term memory (LSTM) encoders, and an attention-based decoder, and the second is based on the Sphinx-4 recognizer, which includes pocket sphinx, base sphinx, and sphinx train, with various types and number of features to be extracted (filter bank and mel frequency cepstral coefficients (MFCC)) based on the CMU Sphinx tool, which generates a language model for different sentences spoken by different speakers. These approaches were tested on a dataset containing 7 hours of spoken Arabic from 11 Arab countries, covering the Levant, Gulf, and African regions, which make up the Arab world, and achieved promising results. CNN-LSTM achieved a word error rate (WER) of 3.63% using 120 features for filter bank and 4.04% WER using 39 features for MFCC, respectively, while the Sphinx-4 recognizer technique achieved 8.17% WER and an accuracy of 91.83% using 25 features for MFCC and 8 Gaussian mixtures, respectively, when tested on the same benchmark dataset.

Study on Real-Time Water Demand Prediction of Winter Wheat–Summer Corn Based on Convolutional Neural Network–Informer Combined Modeling

The accurate prediction of crops’ water requirements is an important reference for real-time irrigation decisions on farmland. In order to achieve precise control of irrigation and improve irrigation water utilization, a real-time crop water requirement prediction model combining convolutional neural networks (CNNs) and the Informer model is presented in this paper, taking the real-time water demand of winter wheat–summer maize from 2017 to 2021 as the research object. The CNN model was used to extract the depth features of the day-by-day meteorological data of the crops, and the extracted feature values were inputted into the Informer model according to the time series for training and prediction to obtain the predicted water demand of winter wheat and summer maize. The results showed that the prediction accuracy of the constructed CNN–Informer combination model was higher compared to CNN, BP, and LSTM models, with an improvement of 1.2%, 25.1%, and 21.9% for winter wheat and 0.4%, 37.4%, and 20.3% for summer maize; based on the good performance of the model in capturing the long-term dependency relationship, the irrigation analysis using the model prediction data showed a significant water-saving effect compared with the traditional irrigation mode, with an average annual water saving of about 1004.3 m3/hm2, or 18.4%, which verified the validity of the model, and it can provide a basis for the prediction of crops’ water demand and sustainable agricultural development.

Pressure prediction for air cyclone centrifugal classifier based on CNN-LSTM enhanced by attention mechanism

In this study, deep learning (DL) is employed to analyze the entire flow field and predict the pressure in air cyclone centrifugal classifiers. To this end, a novel neural network model is proposed, which combines convolutional neural network (CNN), long short-term memory (LSTM) network, and attention mechanism. This study predicts the cases of air cyclone centrifugal classifiers with and without screening cages. Compared to a support vector machine (SVM) and a backpropagation neural network (BPNN) model, the RMSE (root mean square error) of the proposed model with screening cage is reduced by 22.6% and 56.0%, respectively; the RMSE of the proposed model without screening cage is reduced by 50.1% and 42.8%, respectively. This demonstrates the effectiveness and versatility of the CNN-LSTM-Attention model in predicting the pressure of air cyclone centrifugal classifiers. In addition, this study explored the effect of different parameters on air classifier by macroscopic and microscopic screening efficiency. The results show that the air classifier can achieve the best screening effect when the rotational speed is 45 Hz, the feeding speed is 0.3 kg/s and the inclination angle is −4°. This study is expected to provide new ideas for pressure prediction and flow field simulation in air classifiers.

A Hybrid CNN-LSTM Approach for Precision Deepfake Image Detection Based on Transfer Learning

The detection of deepfake images and videos is a critical concern in social communication due to the widespread utilization of deepfake techniques. The prevalence of these methods poses risks to trust and authenticity across various domains, emphasizing the importance of identifying fake faces for security and preventing socio-political issues. In the digital media era, deep learning outperforms traditional image processing methods in deepfake detection, underscoring its significance. This research introduces an innovative approach for detecting deepfake images by employing transfer learning in a hybrid architecture that combines convolutional neural networks (CNNs) and long short-term memory (LSTM). The hybrid CNN-LSTM model exhibits promise in combating deep fakes by merging the spatial awareness of CNNs with the temporal context understanding of LSTMs. Demonstrating effective performance on open-source datasets like “DFDC” and “Ciplab”, the proposed method achieves an impressive precision of 98.21%, indicating its capability to accurately identify deepfake images with a limited false-positive rate. The model’s error rate is 0.26%, emphasizing the challenges and intricacies inherent in deepfake detection tasks. These findings underscore the potential of hybrid deep learning techniques for addressing the urgent issue of deepfake image detection.

Task offloading method based on CNN-LSTM-attention for cloud–edge–end collaboration system

Most of the existing task offloading methods in edge computing environments do not fully utilize the processing capabilities of cloud servers and have high computational complexity, making them unsuitable for real-time processing tasks. To address this problem, we propose a cloud–edge–end collaborative task offloading method based on deep learning methods, combining Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), and attention mechanisms. This method models the total cost of the cloud–edge–end collaboration system as a weighted sum function of the time delay and energy consumption of task execution. Then, with the objective of minimizing the total cost of the system, the task offloading problem is formulated as a mixed-integer joint optimization problem with offloading decision and bandwidth allocation, and two subproblems are decomposed from the optimization problem: one focuses on offloading decision and the other on bandwidth allocation. A CNN-LSTM-Attention neural network-based method is proposed to solve the optimal offloading decision efficiently. Based on this, the differential evolution algorithm is used to generate the optimal bandwidth allocation, resulting in efficient task offloading. The simulation experiment results demonstrate that our method enhances system performance and reduces the overall cost of the system, which is significantly better than existing methods.

Autism spectrum disorder diagnosis with EEG signals using time series maps of brain functional connectivity and a combined CNN–LSTM model

Background and Objective:People with autism spectrum disorder (ASD) often have cognitive impairments. Effective connectivity between different areas of the brain is essential for normal cognition. Electroencephalography (EEG) has been widely used in the detection of neurological diseases. Previous studies on detecting ASD with EEG data have focused on frequency-related features. Most of these studies have augmented data by splitting the dataset into time slices or sliding windows. However, such approaches to data augmentation may cause the testing data to be contaminated by the training data. To solve this problem, this study developed a novel method for detecting ASD with EEG data. Methods:This study quantified the functional connectivity of the subject’s brain from EEG signals and defined the individual to be the unit of analysis. Publicly available EEG data were gathered from 97 and 92 subjects with ASD and typical development (TD), respectively, while they were at rest or performing a task. Time-series maps of brain functional connectivity were constructed, and the data were augmented using a deep convolutional generative adversarial network. In addition, a combined network for ASD detection, based on convolutional neural network (CNN) and long short-term memory (LSTM), was designed and implemented. Results:Based on functional connectivity, the network achieved classification accuracies of 81.08% and 74.55% on resting state and task state data, respectively. In addition, we found that the functional connectivity of ASD differed from TD primarily in the short-distance functional connectivity of the parietal and occipital lobes and in the distant connections from the right temporoparietal junction region to the left posterior temporal lobe. Conclusions:This paper provides a new perspective for better utilizing EEG to understand ASD. The method proposed in our study is expected to be a reliable tool to assist in the diagnosis of ASD.