Search

Advancements in computer vision have resulted in significant breakthroughs across various applications, and one notable area of progress is in the recognition of food ingredient types and states. The identification of food items, distinguishing between types like oranges or apples, and assessing their states, whether whole, peeled, sliced, or juiced, is a pivotal task with far-reaching implications for fields such as food safety, recipe analysis, and restaurant quality control. This paper introduces an innovative approach to food type and state recognition that capitalizes on attention mechanisms and incorporates mask fusion to improve the accuracy and robustness of the recognition process. We evaluate the proposed approach through quantitative and qualitative analyses and comparisons to previous methods. The results consistently demonstrate that our proposed approach, integrating attention mechanisms, outperforms baseline and state-of-the-art methods, achieving an accuracy of 87.11%. This achievement signifies a step forward in refining food image segmentation models and reinforces the applicability of advanced techniques in real-world scenarios. [ABSTRACT FROM AUTHOR]

RNA-protein interaction (RPI) is crucial to the life processes of diverse organisms. Various researchers have identified RPI through long-term and high-cost biological experiments. Although numerous machine learning and deep learning-based methods for predicting RPI currently exist, their robustness and generalizability have significant room for improvement. This study proposes LPI-MFF, an RPI prediction model based on multi-source information fusion, to address these issues. The LPI-MFF employed protein–protein interactions features, sequence features, secondary structure features, and physical and chemical properties as the information sources with the corresponding coding scheme, followed by the random forest algorithm for feature screening. Finally, all information was combined and a classification method based on convolutional neural networks is used. The experimental results of fivefold cross-validation demonstrated that the accuracy of LPI-MFF on RPI1807 and NPInter was 97.60% and 97.67%, respectively. In addition, the accuracy rate on the independent test set RPI1168 was 84.9%, and the accuracy rate on the Mus musculus dataset was 90.91%. Accordingly, LPI-MFF demonstrated greater robustness and generalization than other prevalent RPI prediction methods. [ABSTRACT FROM AUTHOR]

With the rapid development of information technology, the conception of smart healthcare has progressively come to the fore. Smart healthcare utilizes next-generation technologies, such as artificial intelligence, the Internet of Things (IoT), big data and cloud computing to transform intelligently the existing medical system-making it more efficient, more reliable, and personalized. In this work, skin data are collected using dedicated hardware from mobile health units-working as nodes. The collected samples are uploaded to the cloud for further processing using a novel multi-modal information fusion framework, which performs skin lesion segmentation, followed by classification. The proposed framework has two main functional blocks: Segmentation and classification. In each block, we have a performance booster, which works on the principle of information fusion. For lesion segmentation, a hybrid framework is proposed, which utilizes the complementary strengths of two convolutional neural network (CNN) architectures to generate the segmented images. The resultant binary images are later fused using joint probability distribution and marginal distribution function. For lesion classification, a 30-layered CNN architecture is designed, which is trained on the HAM10000 dataset. A novel summation discriminant correlation analysis technique is used to fuse the extracted features from two fully connected layers. To avoid feature redundancy, a feature selection method "Regular Falsi" is developed, which down samples the extracted features into the lower dimensions. The selected features are finally classified using an extreme learning machine classifier. Five skin benchmark datasets (ISBI2016, ISIC2017, ISBI2018, ISIC2019, and HAM10000) are used to evaluate both segmentation and classification frameworks using average accuracy, false-negative rate, sensitivity, and computational time, whose results are impressive compared to state-of-the-art methods. [ABSTRACT FROM AUTHOR]

The convolutional neural network showed considerable success in medical imaging with explainable AI for cancer detection and recognition. However, the irrelevant and large number of features increases the computational time and decreases the accuracy. This work proposes a deep learning and fuzzy entropy slime mould algorithm-based architecture for multiclass skin lesion classification. In the first step, we employed the data augmentation technique to increase the training data and further utilized it for training two fine-tuned deep learning models such as Inception-ResNetV2 and NasNet Mobile. Then, we used transfer learning on augmented datasets to train both models and obtained two feature vectors from newly fine-tuned models. Later, we applied a fuzzy entropy slime mould algorithm on both vectors to get optimal features that are finally fused using the Serial-Threshold fusion technique and classified using several machine learning classifiers. Eventually, the explainable AI technique named Gradcam opted for the visualization of the lesion region. The experimental process was conducted on two datasets, such as HAM10000 and ISIC 2018, and achieved 97.1 and 90.2% accuracy, better than the other techniques. [ABSTRACT FROM AUTHOR]

Moyamoya disease (MMD) is a serious intracranial cerebrovascular disease. Cerebral hemorrhage caused by MMD will bring life risk to patients. Therefore, MMD detection is of great significance in the prevention of cerebral hemorrhage. In order to improve the accuracy of digital subtraction angiography (DSA) in the diagnosis of ischemic MMD, in this paper, a deep network architecture combined with 3D convolutional neural network (3D CNN) and bidirectional convolutional gated recurrent unit (BiConvGRU) is proposed to learn the spatiotemporal features for ischemic MMD detection. Firstly, 2D convolutional neural network (2D CNN) is utilized to extract spatial features for each frame of DSA. Secondly, the long-term spatiotemporal features of DSA sequence are extracted by BiConvGRU. Thirdly, the short-term spatiotemporal features of DSA are further extracted by 3D convolutional neural network (3D CNN). In addition, different features are extracted when gray images and optical flow images pass through the network, and multiple features are extracted by features fusion. Finally, the fused features are utilized to classify. The proposed method was quantitatively evaluated on a data sets of 630 cases. The experimental results showed a detection accuracy of 0.9788, sensitivity and specificity were 0.9780 and 0.9796, respectively, and area under curve (AUC) was 0.9856. Compared with other methods, we can get the highest accuracy and AUC. The experimental results show that the proposed method is stable and reliable for ischemic MMD detection, which provides an option for doctors to accurately diagnose ischemic MMD. [ABSTRACT FROM AUTHOR]

This paper proposes an efficient deep convolutional neural network with features fusion for recognizing radar signal, which mainly includes data pre-processing, features extraction, multi-features fusion, and classification. Radar signals are first transformed into time-frequency images by using choi-williams distribution and smooth pseudo-wigner-ville distribution, and the image pre-processing methods are used to resize and normalize the time-frequency images. Then, two constructed deep convolutional neural network models are aimed to extract more effective features. Furthermore, a multi-features fusion model is proposed to integrate features extracted from two deep convolutional neural network models, which makes full use of the relationship among different features and further improves the recognition performance. Experimental results shown that the average recognition accuracy of the proposed method is up to 84.38% when the signal to noise ratio is at −12 dB, and even reach to 94.31% at −10 dB, which achieved the superior recognition performance than others, especially at the lower signal to noise ratio. Moreover, the recognition performance of various radar signals can be largely improved, especially for 2FSK, 4FSK and SFM. This work provides a sound experimental foundation for further improving radar signal recognition in modern electronic warfare systems. [ABSTRACT FROM AUTHOR]

Breast cancer is diagnosed using mammography imaging. Mammography is an effective screening tool for diagnosing and managing breast cancer. This task is highly time-consuming due to the high similarity between benign and malignant cells. However, for medical intervention, it is important to diagnose breast cancer at an early stage. Recently, deep learning has shown remarkable success in the area of medical imaging for the diagnosis of several cancer types. The DL-based computerized techniques assist in detecting and classifying breast cancer correctly. This article proposes a new computerized architecture based on two novel CNN architectures with Bayesian Optimization and feature selection techniques. Initially, two convolutional neural network (CNN) architectures were designed, named 2-Residual Blocks CNN and 3-Residual Blocks CNN. Both designed architectures have been trained using Mammography images, where hyperparameters have been initialized using Bayesian Optimization (BO). After the training from scratch, deep features are extracted from the average pool layer. Extracted deep features are optimized using an improved optimization algorithm named Simulated Annealing controlled Position Shuffling (SAcPS). The fitness of each iteration is computed using an extreme learning machine (ELM) classifier instead of a fine k-nearest neighbor. The selected features of both CNN architectures are finally fused using a novel serial-controlled Reyni Entropy technique. The fused feature vector is passed to neural networks for final classification. The experimental process was conducted on two publicly available datasets and obtained improved accuracy of 97.7% and 97.3%, respectively. In addition, a detailed comparison is conducted with several recent techniques and shows improved performance. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Computer Engineering & Applications is the property of Beijing Journal of Computer Engineering & Applications Journal Co Ltd. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Knee osteoarthritis (KOA) is one of the deadliest forms of arthritis. If not treated at an early stage, it may lead to knee replacement. That is why early diagnosis of KOA is necessary for better treatment. Manually KOA detection is a time-consuming and error-prone task. Computerized methods play a vital role in accurate and speedy detection. Therefore, the classification and localization of the KOA method are proposed in this work using radiographic images. The two-dimensional radiograph images are converted into three-dimensional and LBP features are extracted having the dimension of N × 59 out of which the best features of N × 55 are selected using PCA. The deep features are also extracted using Alex-Net and Dark-net-53 with the dimensions of N × 1024 and N × 4096, respectively, where N represents the number of images. Then, N × 1000 features are selected individually from both models using PCA. Finally, the extracted features are fused serially with the dimension of N × 2055 and passed to the classifiers on a 10-fold cross-validation that provides an accuracy of 90.6% for the classification of KOA grades. The localization model is proposed with the combination of an open exchange neural network (ONNX) and YOLOv2 that is trained on the selected hyper-parameters. The proposed model provides 0.98 mAP for the localization of classified images. The experimental analysis proves that the presented framework provides better results as compared to existing works. [ABSTRACT FROM AUTHOR]

With the intelligent development of industrial production, sound monitoring technology has been widely used to monitor the operation status of mechanical facilities, and this progress has gradually become a research hotspot in the steel manufacturing industry. However, the complex composition of sound sources and the high computational requirements of most models limit their applicability to industrial scenarios. This work presents a targeted approach for remote monitoring of rolling sound. They are using visual features of audio signals to design classification models on the self-harvested dataset for more efficient adaptation to complex production sites. An in-depth analysis of the actual sound reveals that it is characterized by high similarity, complexity, and partial synchronization. To optimize interclass and intraclass data engineering few-sample, multi-feature fusion, and data augmentation methods were combined to characterize the details fully. In addition, by leveraging the deep small convolutions of VGGNet and the randomness of stochastic pooling, local features are effectively extracted. Finally, global average pooling with final softmax layer stacking is used to classify the rolling signals to reduce the number of parameters, avoid overfitting, and perform a global analysis of the features. Experimental results on the rolling sound dataset show that the method proposed achieves an accuracy of 91.26%, with 92.66%, and 95.88% on the ESC10 and MIMII datasets, respectively. These results confirm that the method can be widely applied to sound classification for multi-category rolling processes, showing good performance and scalability. [ABSTRACT FROM AUTHOR]

Summary: Among all diabetic‐related complications, the diabetic foot ulcer (DFU) is severe, demanding serious attention, and timely treatment. The purpose of the present study was to conduct features fusion of machine learning (ML) based handcrafted low‐level and convolutional neural networks (CNNs) based high‐level features for improving automatic diagnosis of DFU. A standard image dataset containing 1038 abnormal (DFU) and 641 normal color skin patches has been used for experimental evaluation. Handcrafted features are extracted to uncover the edge, color, shape, and texture information from the images. Also, a new CNN architecture has been proposed based on deeper residual blocks for extracting high‐level features. Features fusion with several ML classifiers like logistic regression classifier, support vector machine, gradient boosting, and artificial neural network have shown improved DFU identification results compared to the individual feature categories. Thereby, LRC outperformed the state‐of‐the‐art results for all evaluation metrics by achieving 95.23% sensitivity, 95.37% F1‐score, and 96.50% area under the curve (AUC) value. Such impressive experimental observations suggest that the proposed approach is expected to provide efficient decision support to medical practitioners, thereby improving patient care. [ABSTRACT FROM AUTHOR]

Teledermatology is one of the most illustrious applications of telemedicine and e-health. In this field, telecommunication technologies are utilized to transfer medical information to the experts. Due to the skin's visual nature, teledermatology is an effective tool for the diagnosis of skin lesions especially in rural areas. Furthermore, it can also be useful to limit gratuitous clinical referrals and triage dermatology cases. The objective of this research is to classify the skin lesion image samples, received from different servers. The proposed framework is comprised of two module, which include the skin lesion localization/segmentation and the classification. In the localization module, we propose a hybrid strategy that fuses the binary images generated from the designed 16-layered convolutional neural network model and an improved high dimension contrast transform (HDCT) based saliency segmentation. To utilize maximum information extracted from the binary images, a maximal mutual information method is proposed, which returns the segmented RGB lesion image. In the classification module, a pre-trained DenseNet201 model is re-trained on the segmented lesion images using transfer learning. Afterward, the extracted features from the two fully connected layers are down-sampled using the t-distribution stochastic neighbor embedding (t-SNE) method. These resultant features are finally fused using a multi canonical correlation (MCCA) approach and are passed to a multi-class ELM classifier. Four datasets (i.e., ISBI2016, ISIC2017, PH2, and ISBI2018) are employed for the evaluation of the segmentation task, while HAM10000, the most challenging dataset, is used for the classification task. The experimental results in comparison with the state-of-the-art methods affirm the strength of our proposed framework. [ABSTRACT FROM AUTHOR]

Vision-based human action recognition (HAR) is a hot topic of research from the decade due to a few popular applications such as visual surveillance and robotics. For correct action recognition, various local and global points are requires known as features. These features modified during the variation in human movement. But due to a bit change in several human actions, the features of these actions are mixed that degrade the recognition performance. In this article, we design a new 26-layered Convolutional Neural Network (CNN) architecture for accurate complex action recognition. The features are extracted from the global average pooling layer and fully connected (FC) layer, and fused by a proposed high entropy-based approach. Further, we propose a feature selection method name Poisson distribution along with Univariate Measures (PDaUM). Few of fused CNN features are irrelevant, and few of them are redundant that makes the incorrect prediction among complex human actions. Therefore, the proposed PDaUM based approach selects only the strongest features that later passed to the Extreme Learning Machine (ELM) and Softmax for final recognition. Four datasets are using for experimental analysis - HMDB51 (51 classes), UCF Sports (10 classes), KTH (6 classes), and Weizmann (10 classes). On these datasets, the ELM classifier gives an improved performance as compared to a Softmax classifier. The achieved accuracy on each dataset is 81.4%, 99.2%, 98.3%, and 98.7%, respectively. Comparison with existing techniques, it is shown that the proposed architecture gives better performance in terms of accuracy and testing time. [ABSTRACT FROM AUTHOR]

Although the salient object detection method based on the fully convolutional neural network has achieved better performance, how to learn effective feature representations in complex scenes to obtain more accurate saliency maps is still a challenge. In order to cope with the above-mentioned challenges, an addition or cascade structure is generally used to fuse feature information between multiple levels. However, these methods are susceptible to the influence of messy background information. The network may regard non-salient objects with similar salient appearances as target predictions, and the prediction results may be incomplete due to different appearance areas of salient objects. We design a network composed of multiple attention mechanisms to selectively integrate deep and shallow feature information, and more effectively deal with the transfer and fusion of features. In this paper, we propose a multi-attention embedded network (MAENet), which introduces attention mechanisms to give different feature information with different weights for handling the transfer and aggregation of features at different levels. The multi-attention feature aggregation (MAFA) module is proposed, which uses the channel attention mechanism to give different weights to the features to be fused, and then uses the spatial attention mechanism to selectively aggregate shallow edge information and deep abstract semantic features to avoid excessive redundant information which affects the saliency mapping, as well as suppressing non-salient areas with "salient" appearance. In addition, The multi-scale feature extraction (MFE) module and the self-attention (SA) module are also proposed for obtaining sufficiently rich and useful multi-scale context information and enhancing the function of the top layer. Finally, the attentional residual refinement (ARR) module is utilized to refine the saliency map after each feature fusion and further improve the input function. MAENet can accurately segment salient objects and provide clear local details. Experimental results on five benchmark datasets show that the proposed method achieves the favorable performance against 14 state-of-the-art methods on popular evaluation metrics. [ABSTRACT FROM AUTHOR]

An emerging type of therapeutic agent, anticancer peptides (ACPs), has attracted attention because of its lower risk of toxic side effects. However process of identifying ACPs using experimental methods is both time-consuming and laborious. In this study, we developed a new and efficient algorithm that predicts ACPs by fusing multiview features based on dual-channel deep neural network ensemble model. In the model, one channel used the convolutional neural network CNN to automatically extract the potential spatial features of a sequence. Another channel was used to process and extract more effective features from handcrafted features. Additionally, an effective feature fusion method was explored for the mutual fusion of different features. Finally, we adopted the neural network to predict ACPs based on the fusion features. The performance comparisons across the single and fusion features showed that the fusion of multi-view features could effectively improve the model’s predictive ability. Among these, the fusion of the features extracted by the CNN and composition of k-spaced amino acid group pairs achieved the best performance. To further validate the performance of our model, we compared it with other existing methods using two independent test sets. The results showed that our model’s area under curve was 0.90, which was higher than that of the other existing methods on the first test set and higher than most of the other existing methods on the second test set. The source code and datasets are available at https://github.com/wame-ng/DLFF-ACP. [ABSTRACT FROM AUTHOR]

Timely segmentation and repair of pavement cracks significantly impacts both traffic safety and the service life of the roadway. However, current crack segmentation algorithms are limited in terms of imprecise segmentation and poor robustness. To overcome current limitations, this study proposes a pavement crack segmentation algorithm called MixCrackNet. MixCrackNet leverages deformable convolution, weighted loss functions, an efficient multi-scale attention module, and the Mix Structure to identify pavement cracks. Three datasets were used to train and validate the effectiveness of MixCrackNet. By comparing with classical semantic segmentation networks, the results demonstrate that MixCrackNet outperforms all the other models in crack segmentation. Furthermore, MixCrackNet not only exhibits exceptional performance across all three datasets, but also achieves decent results in untrained dataset. These results indicate that MixCrackNet is not only highly accurate but also robust, thereby promoting the application of semantic crack segmentation technology in pavement condition detection. • We propose a highly accurate and robust model for pavement crack segmentation. • The EMA attention mechanism is used to encode global information and aggregate parallel output feature maps. • The Mix Structure is proposed to fuse shallow and deep features of the ResNet50 network. • We propose an upsampling module based on deformable convolution that can greatly improve crack segmentation accuracy. [ABSTRACT FROM AUTHOR]

Skin cancer is one of the most aggressive cancers in the world. Computer-Aided Diagnosis (CAD) system for cancer detection and classification is a top-rated solution that decreases human effort and time with very high classification accuracy. Machine learning (ML) and deep learning (DL) based approaches have been widely used to develop robust skin-lesion classification systems. Each of the techniques excels when the other fails. Their performances are closely related to the size of the learning dataset. Thus, approaches that are based on the ML are less potent than those found on the DL when working with large datasets and vice versa. In this article, we propose a powerful skin-lesion classification approach based on a fusion of handcrafted features (shape, skeleton, color, and texture) and features extracted from most powerful DL architectures. This combination will make it possible to remedy the limitations of both the ML and DL approaches for the case of large and small datasets. Features engineering is then applied to remove redundant features and to select only relevant features. The proposed approach is validated and tested on both small and large datasets. A comparative study is also conducted to compare the proposed approach with different and recent approaches applied to each dataset. The results obtained show that this features-fusion based approach is very promising and can effectively combine the power of ML and DL based approaches. [ABSTRACT FROM AUTHOR]

• In this research, we introduced a novel TB-UNet based on dilated fusion block (DF) and Attention gate (AG) mechanism for accurate segmentation of lung regions. • We also proposed TB-DenseNet, which is based on five dual convolution blocks, DenseNet-169 layer, and a feature fusion block for the precise classification of tuberculosis images. • The experiments have been performed on three chest X-ray (CXR) datasets, segmented images, and original images are fed to TB-DenseNet for better classification. Tuberculosis still significantly impacts the world's population, with more than 10 million people getting sick each year. Researchers have focused on developing computer-aided diagnosis (CAD) technology based on X-ray imaging to support the identification of tuberculosis, and deep learning is a popular and efficient method. However, deep learning-based CAD approaches usually ignore the relationship between the two vision tasks of specific region segmentation and classification. In this research, we introduced a novel TB-UNet, which is based on dilated fusion block (DF) and Attention block (AB) block for accurate segmentation of lungs regions and achieved the highest results in terms of Precision (0.9574), Recall (0.9512), and F1score (0.8988), IoU (0.8168) and Accuracy (0.9770). We also proposed TB-DenseNet which is based on five dual convolution blocks, DenseNet-169 layer, and a feature fusion block for the precise classification of tuberculosis images. The experiments have been performed on three chest X-ray (CXR) datasets, segmented images, and original images are fed to TB-DenseNet for better classification. Furthermore, the proposed method is tested against simultaneously three different diseases, such as Pneumonia, COVID-19, and Tuberculous. The highest results are achieved in terms of Precision (0.9567), Recall (0.9510), F1score (0.9538), and Accuracy (0.9510). The achieved results reflect that our proposed method produces the highest accuracy compared to the state-of-the-art methods. The source code is available at: https://github.com/ahmedeqbal/TB-DenseNet. [ABSTRACT FROM AUTHOR]

Speech emotion recognition (SER) has become a crucial topic in the field of human–computer interactions. Feature representation plays an important role in SER, but there are still many challenges in feature representation such as the inability to predict which features are most effective for SER and the cultural differences in emotion expression. Most previous studies use a single type of feature for the recognition task or conduct early fusion of features. However, a single type of feature cannot well reflect the emotions of speech signals. Also, different features contain different information, direct fusion cannot integrate the advantages of different features. To overcome these challenges, this paper proposes a parallel network for multi-scale SER based on a connection attention mechanism (AMSNet). AMSNet fuses fine-grained frame-level manual features with coarse-grained utterance-level deep features. Meanwhile, it adopts different speech emotion feature extraction modules according to the temporal and spatial features of speech signals, which enriches features and improves feature characterization. The network consists of a frame-level representation learning module (FRLM) based on the time structure and an utterance-level representation learning module (URLM) based on the global structure. Besides, improved attention-based long short-term memory (LSTM) is introduced into FRLM to focus on the frames that contribute more to the final emotion recognition result. In URLM, a convolutional neural network with the squeeze-and-excitation block (SCNN) is introduced to extract deep features. In addition, the connection attention mechanism is proposed for feature fusion, which applies different weights to different features. Extensive experiments are conducted on the IEMOCAP and EmoDB datasets, and the results demonstrate the effectiveness and performance superiority of AMSNet. Our code will be publicly available at https://codeocean.com/capsule/8636967/tree/v1. • An independent training method is proposed for speech feature training. • Connection attention mechanism is designed for speech feature fusion. • The proposed method is evaluated on two open datasets and achieve good performance. [ABSTRACT FROM AUTHOR]

• Encoder-decoder structure is firstly proposed to efficiently segment hand gestures in complex backgrounds. • DRN and ASPP are combined for solving the problems in segmenting different scales of hand regions. • A simplified decoder module is proposed to reduce the complexity by only using three layers of convolution. • DSC is used in each module to make the network less in computation amount and smaller in size (only 1.85 M). Hand Gesture Recognition (HGR) is widely used in human–computer interaction due to its convenience. However, there are still some challenges in real-world scenarios, such as recognizing hand gestures in the complex backgrounds. To this end, the paper proposes a two-stage HGR system to solve the above issue. Specifically, the first stage performs accurate segmentation to segment the hand from the background. The segmentation network combines dilated residual network, atrous spatial pyramid pooling module and a simplified decoder. The segmentation network can effectively determine hand region even in challenging backgrounds. In the second stage, the double-channel Convolutional Neural Networks (CNNs) are presented to improve the recognition performance. The double-channel CNNs can learn features from the RGB input images and the segmented hand images separately. Experiment results show that the proposed method has an accuracy of 91.17% with the model size of 1.8 MB, both of which are better than other state-of-arts in hand gesture recognition. The method successfully constructed a lightweight model while keeping a high gesture recognition accuracy at final. [ABSTRACT FROM AUTHOR]

Early diagnosis of tuberculosis (TB) is an essential and challenging task to prevent disease, decrease mortality risk, and stop transmission to other people. The chest X-ray (CXR) is the top choice for lung disease screening in clinics because it is cost-effective and easily accessible in most countries. However, manual screening of CXR images is a heavy burden for radiologists, resulting in a high rate of inter-observer variances. Hence, proposing a cost-effective and accurate computer aided diagnosis (CAD) system for TB diagnosis is challenging for researchers. In this research, we proposed an efficient and straightforward deep learning network called TBXNet, which can accurately classify a large number of TB CXR images. The network is based on five dual convolutions blocks with varying filter sizes of 32, 64, 128, 256 and 512, respectively. The dual convolution blocks are fused with a pre-trained layer in the fusion layer of the network. In addition, the pre-trained layer is utilized for transferring pre-trained knowledge into the fusion layer. The proposed TBXNet has achieved an accuracy of 98.98%, and 99.17% on Dataset A and Dataset B, respectively. Furthermore, the generalizability of the proposed work is validated against Dataset C, which is based on normal, tuberculous, pneumonia, and COVID-19 CXR images. The TBXNet has obtained the highest results in Precision (95.67%), Recall (95.10%), F1-score (95.38%), and Accuracy (95.10%), which is comparatively better than all other state-of-the-art methods. [ABSTRACT FROM AUTHOR]

• Effective wind power forecasting promotes power dispatch. • Raw data alone cannot predict wind power well. • Attention mechanism can give weight to physical data. • The long-term trend of the abstract features is more related to the predicted wind power. • Feature fusion can obtain more useful information. Wind power forecasting is an important means to alleviate the pressure of peak and frequency regulation in power systems and improve the acceptance capacity of wind power. However, physical attribute data related to wind power have different effects on its forecasting, and the long-term sequence of original features has redundant information, which makes wind power prediction a daunting challenge. To address these problems, this paper proposes a multi-dimensional extended features fusion model called AMC-LSTM to predict wind power. The Attention Mechanism is utilized to dynamically assign the weight of physical attribute data, which effectively deals with the model's failure to distinguish the difference in importance of input data. Convolutional neural network (CNN) is used for short-term abstract feature extraction to obtain local high-dimensional features, and then Long short-term memory (LSTM) is used to extract the long-term trend of local high-dimensional features, which can effectively reduce the problem of inaccurate prediction caused by the mixing of original data. The extracted temporal features and physical features are fused to predict wind power. Using actual operation data of wind turbine, we verified that the proposed AMC-LSTM hybrid model is capable of integrating multi-scale extended features and providing better performance for short-term wind forecasting. [ABSTRACT FROM AUTHOR]

• A comparative analysis of several deep neural networks architectures for multimodal depression recognition. • An high performing approach for detecting depression in less than 8 seconds of speech segment. • Learning temporal dynamics representations from different modalities for clinical depression recognition. • Comparing different strategies of deep features learning and fusion for clinical depression recognition. • Deep model-level fusion based approach for audiovisual depression recognition. Background and Objective: Major Depressive Disorder is a highly prevalent and disabling mental health condition. Numerous studies explored multimodal fusion systems combining visual, audio, and textual features via deep learning architectures for clinical depression recognition. Yet, no comparative analysis for multimodal depression analysis has been proposed in the literature. Methods: In this paper, an up-to-date literature overview of multimodal depression recognition is presented and an extensive comparative analysis of different deep learning architectures for depression recognition is performed. First, audio features based Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) are studied. Then, early-level and model-level fusion of deep audio features with visual and textual features through LSTM and CNN architectures are investigated. Results: The performance of the proposed architectures using an hold-out strategy on the DAIC-WOZ dataset (80 % training, 10 % validation, 10 % test split) for binary and severity levels of depression recognition is tested. Using this strategy, a set of experiments have been performed and they have demonstrated: (1) LSTM-based audio features perform slightly better than CNN ones with an accuracy of 66.25 % versus 65.60 % for binary depression classes. (2) the model level fusion of deep audio and visual features using LSTM network performed the best with an accuracy of 77.16 % , a precision of 53 % for the depressed class, and a precision of 83 % for the non-depressed class. The given network obtained a normalized Root Mean Square Error (RMSE) of 0.15 for depression severity level prediction. Using a Leave-One-Subject-Out strategy, this network achieved an accuracy of 95.38 % for binary depression detection, and a normalized RMSE of 0.1476 for depression severity level prediction. Our best-performing architecture outperforms all state-of-the-art approaches on DAIC-WOZ dataset. Conclusions: The obtained results show that the proposed LSTM-based surpass the proposed CNN-based architectures allowing to learn temporal dynamics representations of multimodal features. Furthermore, model-level fusion of audio and visual features using an LSTM network leads to the best performance. Our best-performing architecture successfully detects depression using a speech segment of less than 8 seconds, and an average prediction computation time of less than 6 m s ; making it suitable for real-world clinical applications. [ABSTRACT FROM AUTHOR]

In this work, we propose a deep learning framework for the classification of COVID-19 pneumonia infection from normal chest CT scans. In this regard, a 15-layered convolutional neural network architecture is developed which extracts deep features from the selected image samples – collected from the Radiopeadia. Deep features are collected from two different layers, global average pool and fully connected layers, which are later combined using the max-layer detail (MLD) approach. Subsequently, a Correntropy technique is embedded in the main design to select the most discriminant features from the pool of features. One-class kernel extreme learning machine classifier is utilized for the final classification to achieving an average accuracy of 95.1%, and the sensitivity, specificity & precision rate of 95.1%, 95%, & 94% respectively. To further verify our claims, detailed statistical analyses based on standard error mean (SEM) is also provided, which proves the effectiveness of our proposed prediction design. [ABSTRACT FROM AUTHOR]