Search

Gaze is an important non-verbal cue for speculating human's attention, which has been widely employed in many human–computer interaction-based applications. In this paper, we propose an improved Itracker to predict the subject's gaze for a single image frame, as well as employ a many-to-one bidirectional Long Short-Term Memory (bi-LSTM) to fit the temporal information between frames to estimate gaze for video sequence. For single image frame gaze estimation, we improve the conventional Itracker by removing the face-grid and reducing one network branch via concatenating the two-eye region images. Experimental results show that our improved Itracker obtains 11.6% significant improvement over the state-of-the-art methods on MPIIGaze dataset and has robust estimation accuracy for different image resolutions under the premise of greatly reducing network complexity. For video sequence gaze estimation, by employing the bi-LSTM to fit the temporal information between frames, experimental results on EyeDiap dataset further demonstrate 3% accuracy improvement. [ABSTRACT FROM AUTHOR]

Indoor place recognition is a challenging problem because of the hard representation to complicated intra-class variations and inter-class similarities.This paper presents a new indoor place recognition scheme using deep neural network. Traditional representations of indoor place almost utilize image feature to retain the spatial structure without considering the object's semantic characteristics. However, we argue that the attributes, state and relationships of objects are much more helpful in indoor place recognition. In particular, we improve the recognition framework by utilizing Place Descriptors (PDs) in text from to connect different types of place information with their categories. Meanwhile, we analyse the ability of Long Short-Term Memory (LSTM) and Convolutional Neural Network (CNN) for classification in natural language, for which we use them to process the indoor place descriptions. In addition, we improve the robustness of the designed deep neural network by combining a number of effective strategies, i.e. L2-regularization, data normalization, and proper calibration of key parameters. Compared with existing state of the art, the proposed approach achieves well performance of 70.73%, 70.08% and 70.16% of accuracy, precision and recall on Visual Genome database respectively. Meanwhile, the accuracy becomes 98.6% after adding voting mechanics. [ABSTRACT FROM AUTHOR]

Bearings are vital part of rotary machines. A failure of bearing has a negative impact on schedules, production operation and even human casualties. Therefore, in prior achieving fault detection and diagnosis (FDD) of bearing is ensuring the safety and reliable operation of rotating machinery systems. However, there are some challenges of the industrial FDD problems. First, there are unbalanced samples because industrial faults rarely occur. Conse-quently, the labeled data which can refer to failure information are limited in the industry and data augmentation methods are critical pre-processing be-fore training data driven models. Second, due to many learnable parameters in model and data of long sequence, both lead to time delay for FDD. There-fore, this paper proposes various data preprocessing methods and Light-Convolutional Neural Network (LCNN). [ABSTRACT FROM AUTHOR]

The intrinsically low spatial resolution of positron emission tomography (PET) leads to image quality degradation and inaccurate image-based quantitation. Recently developed supervised super-resolution (SR) approaches are of great relevance to PET but require paired low- and high-resolution images for training, which are usually unavailable for clinical datasets. In this paper, we present a self-supervised SR (SSSR) technique for PET based on dual generative adversarial networks (GANs), which precludes the need for paired training data, ensuring wider applicability and adoptability. The SSSR network receives as inputs a low-resolution PET image, a high-resolution anatomical magnetic resonance (MR) image, spatial information (axial and radial coordinates), and a high-dimensional feature set extracted from an auxiliary CNN which is separately-trained in a supervised manner using paired simulation datasets. The network is trained using a loss function which includes two adversarial loss terms, a cycle consistency term, and a total variation penalty on the SR image. We validate the SSSR technique using a clinical neuroimaging dataset. We demonstrate that SSSR is promising in terms of image quality, peak signal-to-noise ratio, structural similarity index, contrast-to-noise ratio, and an additional no-reference metric developed specifically for SR image quality assessment. Comparisons with other SSSR variants suggest that its high performance is largely attributable to simulation guidance. [ABSTRACT FROM AUTHOR]

In multi-label text classification, considering the correlation between labels is an important yet challenging task due to the combination possibility in the label space increasing exponentially. In recent years, neural network models have been widely applied and gradually achieved satisfactory performance in this field. However, existing methods either not model the fully internal correlations among labels or not capture the local and global semantic information of text simultaneously, which somewhat affects the classification results finally. In this paper, we implement a novel model for multi-label classification based on sequence-to-sequence learning, in which two different neural network modules are employed, named encoder and decoder respectively. The encoder uses the convolutional neural network to extract the high-level local sequential semantic, which is combined with the word vector to generate the final text representation through the recurrent neuron network and attention mechanism. The decoder, besides using a recurrent neural network to capture the global label correlation, employs an initialized fully connection layer to capture the correlation between any two different labels. When trained on RCV1-v2, AAPD and Ren-CECps datasets, the proposed model outperforms previous work in main evaluation metrics of hamming loss and micro-F1 score. [ABSTRACT FROM AUTHOR]

• A simple model-free butterfly shape-based detection (BSD) method for control valve stiction. • A relatively higher generalization capability with successful stiction detection rate of 80% on stiction loops and 72.7% on non-stiction loops. • The proposed BSD-CNN quantification algorithm has shown reasonable accuracy when compared to other established methods. Control valve stiction is a long-standing problem within process industries. In most methods for shape-based stiction detection, they rely heavily on the traditional controller output (OP) and process variable (PV) plot (i.e. PV-OP plot) that tends to produce an "elliptical" shape which is the widely acknowledged pattern indication for the presence of stiction. However, many of the methods suffered from unsatisfactory generalization capability when subjected to different loop dynamics. In this paper, a "butterfly" shape derived from the manipulation of the standard PV and OP data, which is more robust towards different loop dynamics, is developed for stiction detection. This simple model-free butterfly shape-based detection (BSD) method uses Stenman's one parameter stiction model, which results in a distinctive 'butterfly' pattern in the presence of stiction. The proposed BSD is tested on simulated data, as well as 26 benchmark industrial case studies and has shown a relatively higher generalization capability with relatively higher successful detection rate on stiction loops and on non-stiction loops. A simple quantification algorithm based on BSD-convolutional neural network (BSD-CNN) framework is then developed to quantify the stiction severity. Based on the 15 benchmark industrial loops with stiction, the proposed BSD-CNN quantification algorithm has shown reasonable accuracy when compared to other published quantification methods in literature. [ABSTRACT FROM AUTHOR]

Authentication plays a very important role to manage security. In the modern era, it is one, in all the priorities. With the appearance of technology, the interaction with machines is turning automatic. Therefore, the need of authentication increases rapidly for various security purposes. Because of this, the biometric-based authentication has gained a drastic momentum. It is a kind of boon over other techniques. However, this event is not a replacement of drawback but varied ways are adopted to verify folks. Signature is one of the first broadly practiced biometric features for the verification of an individual. This paper proposes a method for the pre-processing of signatures to make verification simple. It also proposed a novel method for signature recognition and signature forgery detection with verification using Convolution Neural Network (CNN), Crest-Trough method and SURF algorithm & Harris corner detection algorithm. The proposed system attains an accuracy of 85-89% for forgery detection and 90-94% for signature recognition. [ABSTRACT FROM AUTHOR]

• A novel polycrystalline silicon wafer defect segmentation method based on CNNs is proposed. • The proposed method alleviates unbalanced pixel distribution by adopting region proposal network. • The experimental results on real defect images show that the proposed methods outperforms state-of-the-art methods. Defect segmentation is an important way for defect detection in machine vision. For polycrystalline silicon wafer production, it is difficult to automatically segment defects due to its inhomogeneous background and unpredictable defect shapes. The conventional methods based on handcrafted models or features not only heavily rely on the expertise, but also are not flexible from one application case to another. In this paper, we propose a defect segmentation method for polycrystalline silicon wafer based on the deep convolutional networks. Firstly, we apply Region Proposal Network (RPN) to generate underlying defect regions. Then, these generated image patches are processed to suitable sizes for feeding into a improved segmentation network which is modified based on U-net and a dilation convolution. Real defect images are used to test the proposed method and experimental results show that proposed method achieves better performance compared with the state-of-the-art methods. [ABSTRACT FROM AUTHOR]

• A joint spatial-spectral hyperspectral image classification framework based on convolutional neural network is proposed. • A two-stream convolutional network is designed to learn the spatial-spectral features at different scales. • A spatial enhancement strategy is proposed to obtain more comprehensive spatial features with limited training samples. Hyperspectral image (HSI) classification technology has been widely used in many earth observation tasks, such as detection, recognition, and surveillance. The traditional hyperspectral image classification methods mainly utilize hand-crafted features, such as edge and texture descriptors, which are not robust for different input data. By contrast, deep learning based methods exploit high-level features for hyperspectral image classification, but they usually degenerate the spatial-spectral structure, depend on a large number of training samples, and ignore a large amount of implicitly useful information. To address these problems, a new joint spatial-spectral hyperspectral image classification method based on different-scale two-stream convolutional network and spatial enhancement strategy is proposed in this paper. First, the pixel blocks at different scales around the center pixel are selected as the basic units to be processed. Then, a spatial enhancement strategy is designed to obtain various spatial location information under the limited training samples by the spatial rotation and row-column transformation. Finally, the spatial-spectral feature is learned by a different-scale two-stream convolutional network, and the classification result of the center pixel is obtained by a softmax layer. Experimental results on two datasets demonstrate that the proposed method outperforms other state-of-the-art methods qualitatively and quantitatively. [ABSTRACT FROM AUTHOR]

An automated method for classification of multispectral satellite images using image fusion is presented in this paper. Principal Component Analysis(PCA),and convolutional neural network (CNN).This proposed method first take multispectral satellite image and apply Brovey transform method for fusion of three RGB bands with panchromatic band shaving spatial resolution of 30m and 15m respectively. Then Principal Component Analysis is applied on fused images for reducing dimension. After that Convolutional Neural Network has been applied for classification purpose. Classification outcomes of the proposed method are matched with FCM, GIFP-FCM,and FKLICM. [ABSTRACT FROM AUTHOR]

Continues growth of social networking web users, people daily shared their ideas and opinions in the form of texts, images, videos, and speech. Text categorization is still a crucial issue because these huge texts received from the heterogeneous sources and different mindset peoples. The shared opinion is to be incomplete, inconsistent, noisy and also in different languages form. Currently, NLP and deep neural network methods are widely used to solve such issues. In this way, Word2Vec word embedding and Convolutional Neural Network (CNN) method have to be implemented for effective text classification. In this paper, the proposed model perfectly cleaned the data and generates word vectors from pre-trained Word2Vec model and use CNN layer to extract better features for short sentences categorization. [ABSTRACT FROM AUTHOR]

Automatic electrocardiogram beat classification is a prevalent field of research among the biomedical and computer science researchers. In the last decade, the research has seen a transition from feature centric classification approach to a classifier centric approach. The performance of conventional approach based classifiers such as probabilistic neural networks, support vector machines, is highly dependent on the performance of pre-processing and feature extraction modules. Whereas, modern approach based classifiers such as deep neural networks, convolutional neural networks, do not need handcrafted features. They can generate characteristic features automatically from raw ECG signal. This paper aims to provide a comparative analysis of conventional and modern approaches. Further, deep learning based classifier models have been investigated to design a road-map for researcher new to this field. [ABSTRACT FROM AUTHOR]

Online reviews, social media data contains wealth of information about customer's likes and dislikes about product or service. Extreme opinions in terms of praises and complaints are more informative as compared to plain positive and plain negative sentences and are difficult to find from customer reviews. Existing approaches to sentiment analysis classify the sentences as positive negative or neutral and do not consider informativeness of sentences. This paper compares Machine Learning, Ensemble and Deep Learning based approaches for praise or complaint classification using hybrid features based on linguistic properties of extreme opinions. The performance of the Machine Learning classifiers with proposed hybrid features and Deep Learning based classifiers using Dense Neural Network, CNN, Multichannel CNN were evaluated by training and testing the deep neural network with a set of important words such as nouns, adjectives, intensifiers, verbs, etc present in the sentence. The proposed Ensemble classifier uses SVC, Random Forest, Decision Tree, MLP classifiers. Hotel domain reviews were evaluated using the parameters Accuracy, Precision, Recall, and F1-score. The evaluation results show that the proposed Ensemble classifier and Deep Learning Models provides better classification accuracy as compared to existing methods. [ABSTRACT FROM AUTHOR]

Image fusion using image normalization and radiometric calibration and Particle Swarm Optimization Fuzzy-c means (PSOFCM) based unsupervised change observation technique is proposed in this paper. To obtain a change of an image, principle component analysis (PCA) & absolute difference log ratio images are fused using radiometric calibration and followed by PSOFCM classifier. Further PSOFCM classifier is used for classifying the satellite image segments into change & unchanged blocks. Better performance is obtained from the simulation of proposed method as compare to other states of art methods. [ABSTRACT FROM AUTHOR]

Corn is one of the most popular food grains in the India and crop loss due to diseases substantially affects the Indian economy and threatens the food availability. Recent access of smart devices can be utilized to provide automatic diagnosis of corn diseases and prevent severe crop losses. This paper presents a real time method based on deep convolutional neural network for corn leaf disease recognition. Deep neural network performance is improved by tuning the hyper-parameters and adjusting the pooling combinations on a system with GPU. Further, the number of parameters of the developed model is optimized to make it suitable for real time inference. The pre-trained deep CNN model was deployed onto raspberry pi 3 using Intel Movidius Neural Compute Stick consisting dedicated CNN hardware blocks. During the recognition of corn leaf diseases, the deep learning model achieves an accuracy of 88.46% demonstrating the feasibility of this method. The presented corn plant disease recognition model is capable of running on standalone smart devices like raspberry-pi or smart-phone and drones. [ABSTRACT FROM AUTHOR]

Deep convolutional neural networks (CNNs) have attracted great attention in the field of image denoising. However, there are two drawbacks: (1) it is very difficult to train a deeper CNN for denoising tasks, and (2) most of deeper CNNs suffer from performance saturation. In this paper, we report the design of a novel network called a batch-renormalization denoising network (BRDNet). Specifically, we combine two networks to increase the width of the network, and thus obtain more features. Because batch renormalization is fused into BRDNet, we can address the internal covariate shift and small mini-batch problems. Residual learning is also adopted in a holistic way to facilitate the network training. Dilated convolutions are exploited to extract more information for denoising tasks. Extensive experimental results show that BRDNet outperforms state-of-the-art image-denoising methods. The code of BRDNet is accessible at http://www.yongxu.org/lunwen.html. [ABSTRACT FROM AUTHOR]

Deep convolutional neural networks (CNNs) are the dominant technology in computer vision today. Much of the recent computer vision literature can be thought of as a competition to find the best architecture for vision within the deep convolutional framework. Despite all the effort invested in developing sophisticated convolutional architectures, however, it's not clear how different from each other the best CNNs really are. This paper measures the similarity between two well-known CNNs, Inception and ResNet, in terms of the properties they extract from images. We find that the properties extracted by Inception are very similar to the properties extracted by ResNet, in the sense that either feature set can be well approximated by an affine transformation of the other. In particular, we find evidence that the information extracted from images by ResNet is also extracted by Inception, and in some cases may be more robustly extracted by Inception. In the other direction, most but not all of the information extracted by Inception is also extracted by ResNet. The similarity between Inception and ResNet features is surprising. Convolutional neural networks learn complex non-linear features of images, and the architectural differences between systems suggest that these non-linear functions should take different forms. Nonetheless, Inception and ResNet were trained on the same data set and seem to have learned to extract similar properties from images. In essence, their training algorithms hill-climb in totally different spaces, but find similar solutions. This suggests that for CNNs, the selection of the training set may be more important than the selection of the convolutional architecture. keyword: ResNet, Inception, CNN, Feature Evaluation, Feature Mapping. [ABSTRACT FROM AUTHOR]

This paper presents a real-time 3D object detector based on LiDAR based Simultaneous Localization and Mapping (LiDAR-SLAM). The 3D point clouds acquired by mobile LiDAR systems, within the environment of buildings, are usually highly sparse, irregularly distributed, and often contain occlusion and structural ambiguity. Existing 3D object detection methods based on Convolutional Neural Networks (CNNs) rely heavily on both the stability of the 3D features and a large amount of labelling. A key challenge is efficient detection of 3D objects in point clouds of large-scale building environments without pre-training the 3D CNN model. To project image-based object detection results and LiDAR-SLAM results onto a 3D probability map, we combine visual and range information into a frustum-based probabilistic framework. As such, we solve the sparse and noise problem in LiDAR-SLAM data, in which any point cloud descriptor can hardly be applied. The 3D object detection results, obtained using both backpack LiDAR dataset and the well-known KITTI Vision Benchmark Suite, show that our method outperforms the state-of-the-art methods for object localization and bounding box estimation. [ABSTRACT FROM AUTHOR]

• We propose a novel video captioning model with global-local attention. • We combine LSTM and 1D CNN in the decoder. • Our model outperforms the state-of-the-art on MSVD and MSR-VTT. Video captioning with encoder–decoder structures has been extensively studied in the recent literature, where a great deal of work focuses on multimodal features and attention mechanisms. Most of the previous work uses only the global temporal features, such as image, motion, and audio features, and ignores the local semantic features existing extensively in the video data. Furthermore, it is difficult to fully utilize the local features due to frame-to-frame redundancy. In this paper, we propose to combine global temporal features and local object-based features in a complementary way to develop a multimodal attention mechanism (global–local attention mechanism). Based on this attention mechanism, we introduce a novel video captioning method, Recurrent Convolutional Video Captioning with Global and Local Attention (RCGL). Further, both LSTM and 1D CNN are incorporated into the decoder to improve the long-range dependency. The experimental results on two standard datasets, MSVD and MSR-VTT, demonstrate that RCGL outperforms the state-of-the-art in four common metrics. [ABSTRACT FROM AUTHOR]

This paper presents an open-source pipeline to train neural networks to segment structures of interest from MRI data. The pipeline is tailored towards homogeneous datasets and requires relatively low amounts of manual segmentations (few dozen, or less depending on the homogeneity of the dataset). Two use-case scenarios for segmenting the spinal cord white and grey matter are presented: one in marmosets with variable numbers of lesions, and the other in the publicly available human grey matter segmentation challenge [ 1 ]. The pipeline is freely available at: https://github.com/neuropoly/multiclass-segmentation. [ABSTRACT FROM AUTHOR]

• A framework for automatic organization and classification of educational videos. • Collection of a large-scale MOOC dataset. • Thorough investigation of data representation techniques for video lecture classification. • Performance evaluation of deep and conventional classifiers on MOOC dataset. The advent of MOOC platforms brought an abundance of video educational content that made the selection of best fitting content for a specific topic a lengthy process. To tackle this challenge in this paper we report our research efforts of using deep learning techniques for managing and classifying educational content for various search and retrieval applications in order to provide a more personalized learning experience. In this regard, we propose a framework which takes advantages of feature representations and deep learning for classifying video lectures in a MOOC setting to aid effective search and retrieval. The framework consists of three main modules. The first module called pre-processing concerns with video-to-text conversion. The second module is transcript representation which represents text in lecture transcripts into vector space by exploiting different representation techniques including bag-of-words, embeddings, transfer learning, and topic modeling. The final module covers classifiers whose aim is to label video lectures into the appropriate categories. Two deep learning models, namely feed-forward deep neural network (DNN) and convolutional neural network (CNN) are examined as part of the classifier module. Multiple simulations are carried out on a large-scale real dataset using various feature representations and classification techniques to test and validate the proposed framework. [ABSTRACT FROM AUTHOR]

Effective motion estimation is one of the prime steps for any human action recognition (HAR) algorithm. Optical flow (OF) and motion history image (MHI) are two well-known methods for motion estimation in videos. OF has several advantages over MHI. But the major drawback with OF is that it is computationally very expensive as compared to the MHI. Therefore, in this paper, a new motion estimation technique named as Weber Motion History Image (WMHI) is proposed. Here, an extremely fast algorithm is proposed for HAR using WMHI, pose information, and convolutional neural network (CNN). In spite of being fast and less space consuming, the algorithm outperforms the existing pose based CNN results on five benchmark datasets namely JHMDB [1], sub-JHMDB [1] , MPII [2] and HMDB51 [3] and UCF101 [4]. The work mainly focuses on a new efficient algorithm which can be implemented for real-time HAR in videos. For real-time implementation, the two basic criteria on which an algorithm can be analyzed are space and time complexity. The proposed algorithm is faster as compared to the existing OF based HAR systems. In terms of space complexity, the feature size of the proposed algorithm is almost 50% of the existing OF based algorithm. The recognition results still outperform the existing result by a significant margin. [ABSTRACT FROM AUTHOR]

Adversarial methods have demonsterated to be signifiant at generating realistic images. However, these approaches have a challenging training process which partially attributed to the performance of discriminator. In this paper, we proposed an efficient super-resolution model based on generative adversarial network (GAN), to effectively generate reprehensive information and improve the test quality of the real-world images. To overcome the current issues, we designed the discriminator of our model based on the Least Square Loss function. The proposed network is organized by a gradual learning process from simple to advanced, which means from the small upsampling factors to the large upsampling factor that helps to improve the overall stability of the training. In particular, to control the model parameters and mitigate the training difficulties, dense residual learning strategy is adopted. Indeed, the key idea of proposed methodology is (i) fully exploit all the image details without losing information by gradually increases the task of discriminator, where the output of each layer is gradually improved in the next layer. In this way the model efficiently generates a super-resolution image even up to high scaling factors (e.g. × 8). (ii) The model is stable during the learning process, as we use least square loss instead of cross-entropy. In addition, the effects of different objective function on training stability are compared. To evaluate the model we conducted two sets of experiments, by using the proposed gradual GAN and the regular GAN to demonstrate the efficiency and stability of the proposed model for both quantitative and qualitative benchmarks. [ABSTRACT FROM AUTHOR]

• This paper proposes a fully convolutional deep architecture for predicting depth and surface normal from a single RGB image. • A state-of-the-art CNN, DenseNet, is employed to extract the deep convolutional features from the input RGB images. • Deconvolution is applied at various stages of the network to enhance the spatial resolution of intermediate feature maps. • Bottleneck layers are incorporated to acquire some feature maps of same resolution as that of the deconvolution blocks. • Concatenated features are progressed along the deep network in a cascaded manner to yield the target (depth, surface normal). Understanding the 3D perspective of a scene is imperative in improving the precision of intelligent autonomous systems. The difficulty in understanding is compounded when only one image of the scene is available at disposal. In this regard, we propose a fully convolutional deep framework for predicting the depth map and surface normal from a single RGB image in a common architecture. The proposed model accepts an input RGB image of size 320 × 240 × 3, and predicts the output via 4 stages: (1) It forwards the input through the convolution layers of DenseNet-161 to extract deep features of size 10 × 7 × 2208, (2) Few deconvolution blocks are appended at the end of the DenseNet in order to enhance the resolution of target output, (3) The output feature maps of the bottleneck layers, which are appended to the dense blocks of the DenseNet to facilitate feature-reuse, are concatenated with same resolution feature maps of deconvolution blocks, (4) These concatenated features are progressed along the network in a cascaded manner to form the final output of resolution 160 × 120. Understanding the 3D perspective of a scene is imperative in improving the precision of intelligent autonomous systems. The difficulty in understanding is compounded when only one image of the scene is available at disposal. In this regard, we propose a fully convolutional deep framework for predicting the depth map and surface normal from a single RGB image in a common architecture. The DenseNet CNN architecture is employed to learn the complex mapping between an input RGB image and its corresponding 3D primitives. We introduce a novel approach of multi-stage cascaded deconvolution, where the output feature maps of one dense block are reused by concatenating with the feature maps of the corresponding deconvolution block. These combined feature maps are progressed along the deep network in a pre-activated manner to construct the final output. The network is trained separately for estimating depth and surface normal while keeping the architecture same. The suggested architecture, compared to the counterparts, uses fewer training samples and model parameters. Exhaustive experiments on benchmark dataset not only reveal the efficacy of the proposed multi-stage scheme over the one-way sequential deconvolution but also outperform the state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Ensuring a healthier working environment is of utmost importance for companies and global health organizations. In manufacturing plants, the ergonomic assessment of adopted working postures is indispensable to avoid risk factors of work-related musculoskeletal disorders. This process receives high research interest and requires extracting plausible postural information as a preliminary step. This paper presents a semi-automated end-to-end ergonomic assessment system of adopted working postures. The proposed system analyzes the human posture holistically, does not rely on any attached markers, uses low cost depth technologies and leverages the state-of-the-art deep learning techniques. In particular, we train a deep convolutional neural network to analyze the articulated posture and predict body joint angles from a single depth image. The proposed method relies on learning from synthetic training images to allow simulating several physical tasks, different body shapes and rendering parameters and obtaining a highly generalizable model. The corresponding ground truth joint angles have been generated using a novel inverse kinematics modeling stage. We validated the proposed system in real environments and achieved a joint angle mean absolute error (MAE) of 3.19±1.57∘ and a rapid upper limb assessment (RULA) grand score prediction accuracy of 89% with Kappa index of 0.71 which means substantial agreement with reference scores. This work facilities evaluating several ergonomic assessment metrics as it provides direct access to necessary postural information overcoming the need for computationally expensive post-processing operations. [ABSTRACT FROM AUTHOR]

Fine-grained semantic segmentation results are typically difficult to obtain for subdecimeter aerial imagery segmentation as a result of complex remote sensing content and optical conditions. Recently, convolutional neural networks (CNNs) have shown outstanding performance on this task. Although many deep neural network structures and techniques have been applied to improve accuracy, few have attended to improving the differentiation of easily confused classes. In this paper, we propose TreeUNet, a tool that uses an adaptive network to increase the classification rate at the pixel level. Specifically, based on a deep semantic model infrastructure, a Tree-CNN block in which each node represents a ResNeXt unit is constructed adaptively in accordance with the confusion matrix and the proposed TreeCutting algorithm. By transmitting feature maps through concatenating connections, the Tree-CNN block fuses multiscale features and learns best weights for the model. In experiments on the ISPRS two-dimensional Vaihingen and Potsdam semantic labelling datasets, the results obtained by TreeUNet are competitive among published state-of-the-art methods. Detailed comparison and analysis show that the improvement brought by the adaptive Tree-CNN block is significant. [ABSTRACT FROM AUTHOR]

Accurate and efficient localization of hydrogen leakage is crucial for ensuring the safe and stable operation of hydrogen refueling stations. In this paper, a hybrid model (CEEMDAN–CNN–LSTM) based on data noise reduction and deep learning is proposed to predict the location of hydrogen leakage in hydrogen refueling stations. The model employs the complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) algorithm to effectively reduce noise and process 720 hydrogen leakage condition datasets. The CNN-LSTM model is constructed by synergistically combining the spatiotemporal feature extraction capabilities of the convolutional neural network (CNN) with the temporal processing capabilities of the long-short-term memory neural network (LSTM) while optimizing hyperparameter combinations through GridSearch. Compared with other models, the proposed CEEMDAN–CNN–LSTM model demonstrates superior prediction performance, achieving an accuracy of 99.54 %, precision of 99.42 %, recall of 99.54 %, and F1-score of 99.46 %. Moreover, when utilizing a 5-s duration of leakage data as input, the model attains an impressive accuracy of 97.64 %. These results underscore the favorable application prospects of the CEEMDAN–CNN–LSTM model in the field of gas leak localization. • A CEEMDAN–CNN–LSTM model is proposed for predicting hydrogen leakage locations. • A leakage database is constructed using CFD numerical simulations. • CEEMDAN is utilized to reduce noise and improve the performance of the model. • CNN is used for feature extraction, and LSTM for temporal prediction. • The accuracy of leakage source localization at hydrogen refueling stations is 99.54 %. [ABSTRACT FROM AUTHOR]

Rapid detection and diagnosis of plant diseases is crucial for healthy crop growth and maintaining harvest quality. Many diseases, such as bacterial spot, late blight, and leaf mould, can affect tomato cultivation. In this paper, a mobile app-based system for intelligent disease detection on tomato leaves is developed using convolutional neural networks (CNNs). The study uses transfer learning approaches to fine-tune well-known CNN architectures like AlexNet, ResNet-50, SqueezeNet-1.1, VGG19, and DenseNet-121 on a dataset made up of images of tomato leaf from the PlantVillage dataset as well as self-captured images. Over 95% accuracy is attained by all trained CNN models, with DenseNet-121 leading the pack with a stellar 99.85%. Consequently, DenseNet-121 is proposed for use with a user-friendly mobile application to detect disease in tomato leaves. A robust user interface in English and two important Indian languages makes this application more accessible to a wider user base. [Display omitted] • This study incorporates real-time field data along with the PlantVillage dataset. • This study fine-tunes 5 CNNs, with DenseNet-121 leading the way with 99.85% accuracy. • The app offers English and 2 Indian languages for broad accessibility and utility. [ABSTRACT FROM AUTHOR]

Among the existing deep learning-based person re-identification (ReID) methods, human parsing based on semantic segmentation is the most promising solution for ReID because such models can learn to identify fine-grained details of different body parts or apparel of a target semantically. However, intra-class variations such as illumination changes, multi-pose angles, and cloth-changing (CC) across different non-overlapping camera viewpoints present a crucial challenge for this approach. Among these challenges, a person CC is the most distinctive problem for ReID models, which often fail to associate the target in new cloth against the learned feature semantics of the previous cloth worn in a different timeline. In this paper, we propose a face and body-shape integration (FBI) network as a tactical solution to address the long-term person CC-ReID problem. The FBI comprises hierarchically stacked parsing and edge prediction (PEP) CNN blocks that generate fine-grained human-parsing output at the initial stage. We then aligned the PEP to our proposed model agnostic plug-in feature overlay module (FOM) to mask cloth-relevant body attributes except the facial features pooled from the input sample. Thus, our human parsing PEP and FOM modules are attuned to discriminatively learn cloth-irrelevant features of the target pedestrian(s) to optimize the effectiveness of person ReID in solitary or minimally crowded areas. In our extensive person CC-ReID experiments, our FBI model achieves 83.4/61.8 in R1 and 91.7/65.8 in mAP evaluation results on the PRCC and LTCC datasets, respectively; thereby significantly out-competing several previous state-of-the-art ReID methods, and validating the effectiveness of the FBI. • FBI model uses a combination of face and body shape features for effective CC-ReID. • We are the first to integrate face features and body contour for long-term CC-ReID. • FBI shows consistent and accurate CC-ReID in indoor and outdoor real-world scenarios. • FBI's R1 & mAP outperforms state-of-the-art CC-ReID methods on PRCC & LTCC datasets. • FBI is a robust model-agnostic method of addressing long-standing CC-ReID problems. [ABSTRACT FROM AUTHOR]

Convolutional neural networks (CNNs) have performed well in image denoising. Although some CNNs enlarge convolutional kernels and increase stacked convolutional layers to overcome the locality defect of convolutional operations, they may increase computational costs. In this paper, we propose a parallel and serial denoising network (PSDNet) for image denoising to preserve image texture. Specifically, the proposed PSDNet contains a parallel block (PB), a serial block (SB), and a reconstruction block (RB). A PB uses two heterogeneous sub-networks with a deformable convolution in a parallel way to extract comparative information for better-recovering image texture. A SB utilizes an enhanced residual dense architecture via combinations of a batch normalization, ReLU, and convolutional layer in a serial way to refine obtained features for obtaining more accurate noise information. A RB is responsible for reconstructing images. Experimental results reveal that our PSDNet is very effective in image denoising, according to quantitative analysis and visual analysis. Codes can be obtained at https://github.com/hellloxiaotian/PSDNet. • Heterogeneous architecture with deformable convolution can better filter noise. • An enhanced residual architecture is used to remove redundant features. • Combining a parallel and serial way can improve effects of images denoising. • Proposed network is effective for synthesized and real noisy image denoising. [ABSTRACT FROM AUTHOR]

This paper proposes a novel data processing scheme for electronic noses that combines the gramian angular field (GAF) and convolutional neural network (CNN) to achieve high performance in classifying five levels of odor intensity. Specifically, a multi-channel e-nose was developed to detect various gases, including hydrogen sulfide, ammonia, sulfur dioxide, trimethylamine, and alkane gases, among others, in complex odor components. The sensor array was optimized through Spearman correlation analysis of the sensor signals and artificial olfactory odor intensity levels. Moreover, the one-dimensional temporal sensor data was converted into two-dimensional color images using the GAF (GASF/GADF) algorithm. This approach enables a more detailed presentation of deep features while retaining the time-domain dependence of the signals. To enhance the performance of classification, a multi-scale feature fusion network (MFFNet) was designed. Notably, GASF-converted images are more effective in characterizing sensor data for different odor intensity levels than GADF-converted images. Compared to classical CNN classification models such as AlexNet, GoogLeNet, and ResNet18, MFFNet achieved the highest accuracy and macro average F1-score on the test set, which were 93.75% and 93.34%, respectively. The results demonstrate the efficient classification of odor intensity levels by combining multi-channel e-nose technology, GASF, and CNN. • The combination of GAF (GASF/GADF) and CNN is introduced into the field of electronic nose data processing. • An efficient prototype of an e-nose system called Hypernose with 30 gas sensors has been independently developed. • A multi-scale feature fusion network (MFFNet) is designed to effectively classify odor intensity levels. [ABSTRACT FROM AUTHOR]

Remaining Useful Life (RUL) prediction of drilling pumps, pivotal components in fossil energy production, is essential for efficient maintenance and safe operation of such facilities. This paper introduces a deep feature learning method that combines a Convolutional Neural Network (CNN)-Convolutional Block Attention Module (CBAM) and a Transformer network into a parallel channel method to predict the RUL of drilling pumps. Specifically, two parallel channels independently extract time-frequency domain and time-domain features from strain signals and then proceed with degradation estimation through feature learning. The deep features derived independently from the two channels are subsequently amalgamated to predict the RUL of the drilling pump. The proposed method is validated by the operational data from four operating drilling pumps. The comparative analysis confirms the higher accuracy of the proposed method over several existing state-of-the-art approaches. Overall, the proposed method supports the safe and cost-saving-oriented operation and maintenance of drilling pumps. • Time and time-frequency domain features are incorporated as inputs of the model. • Feature weights are adaptively assigned by the CBAM and multi-head attention mechanisms. • A parallel feature extraction structure is created for model generalization and overfitting risk reduction. [ABSTRACT FROM AUTHOR]

• Improving the results of Siamese networks using CP-CVV. • Improving One-Shot-Learning results by class partitioning. • Improving the results of CIFAR-FS with CP-CVV. • Tuning CP-CVV hyper-parameters for optimal results. • CP-CVV as an alternative to traditional ensemble models. One Shot Learning includes all those techniques that make it possible to classify images using a single image per category. One of its possible applications is the identification of food products. For a grocery store, it is interesting to record a single image of each product and be able to recognise it again from other images, such as photos taken by customers. Within deep learning, Siamese neural networks are able to verify whether two images belong to the same category or not. In this paper, a new Siamese network training technique, called CP-CVV, is presented. It uses the combination of different models trained with different classes. The separation of validation classes has been done in such a way that each of the combined models is different in order to avoid overfitting with respect to the validation. Unlike normal training, the test images belong to classes that have not previously been used in training, allowing the model to work on new categories, of which only one image exists. Different backbones have been evaluated in the Siamese composition, but also the integration of multiple models with different backbones. The results show that the model improves on previous works and allows the classification problem to be solved, an additional step towards the use of Siamese networks. To the best of our knowledge, there is no existing work that has proposed integrating Siamese neural networks using a class-based validation set separation technique so as to be better at generalising for unknown classes. Additionally, we have applied Cross-Validation-Voting with ConvNeXt to improve the existing classification results of a well-known Grocery Store Dataset. [ABSTRACT FROM AUTHOR]

Human activity recognition (HAR) is an essential task in many applications such as health monitoring, rehabilitation, and sports training. Sensor-based HAR has received increasing attention due to the widespread availability of sensors in daily life. In this paper, we propose two novel architectures, the convolution with self-attention network (CSNet) and the temporal-channel convolution with self-attention network (TCCSNet), for classifying sequences of human activity data from different sensors. CSNet leverages both convolution and self-attention to capture both local and global dependencies in the input data, while TCCSNet exploits both temporal and inter-channel dependencies through two branches of convolutions and self-attentions for extracting time-wise and channel-wise information. The proposed methods are evaluated on seven different sensor-based HAR datasets, namely: MHEALTH, PAMAP2, UTD1, UTD2, WHARF, USC-HAD, and WISDM, using the leave-one-subject-out cross-validation protocol. Our experiments show that the proposed models outperform other modern approaches, such as Transformers and long short-term memory (LSTM) based models. • Novel CSNet and TCCSNet models leverage self-attention for activity recognition. • CSNet captures local patterns & long-range dependencies with convolution & attention. • TCCSNet captures temporal and inter-channel dependencies for enhanced recognition. • Proposed methods outperform state-of-the-art using LOSO protocol on seven datasets. [ABSTRACT FROM AUTHOR]

[Display omitted] • A novel IEDN-RNET model is proposed for the short and medium term PV and wind power forecasting. • The proposed model effectively captures complex and nonlinear temporal information and patterns. • The proposed model improves the PV and wind power forecasting results with an average MAE of 16%. • Natal dataset and CSG diverse datasets are taken to verify the robustness of the proposed model. • The evaluation metrics demonstrate that the IEDN-RNET model achieves the highest accuracy. Accurate and consistent forecasting of regional wind power is essential for efficient scheduling and maximizing the utilization of renewable energy in the power grid. Medium-term PV and wind power forecasting play a vital role in various aspects of renewable energy planning and management. It provides valuable insights into future wind power generation trends, enabling effective resource allocation, grid integration, and energy market operations. This paper proposed a hybrid inception embedded deep neural network with ResNet architecture, Proposed model utilizes ResNet blocks and Inception modules to enhance feature extraction and attention mechanisms. Furthermore, The bidirectional weighted LSTM and GRU layers are used to further increase the model's ability to capture temporal dependencies and provide accurate power generation. The Time2Vec method is integrated to capture important complex and nonlinear temporal information and patterns in the time series data. Our model utilizes both convolutional and recurrent neural network components to learn and predict power generation with improved accuracy. This study employs Chinese State Grid PV and wind datasets and Natal wind datasets to determine the performance of our proposed IEDN-RNET model. The results of our study show that the proposed IEDN-RNET model outperforms other algorithms with a 12% lower Mean Absolute Error, 13% lower Root Mean Square Error, 19% lower Normalized Mean Absolute Error, 20% lower Normalized Root Mean Square Error, and higher R-Square and correlation coefficients. [ABSTRACT FROM AUTHOR]

Non-orthogonal multiple access (NOMA) techniques have drawn much attention for massive connectivity, heterogeneous data traffic with ultra-low latency requirements, and ultra-high bandwidth efficiency for the future generation of wireless communication systems. This paper suggests a deep convolutional-long short-term memory (ConvNet-LSTM) aided receiver for detecting signals in the downlink NOMA system. The proposed technique implicitly estimates channel state information and recovers the transmitted symbol directly. The Bayesian approach optimizes the hyperparameters for the deep ConvNet-LSTM. The proposed learning approach is significantly more efficient than the conventional signal detection method, successive interference cancellation (SIC)-based minimum mean square error (MMSE) technique, other deep learning (DL)-based approaches, long short-term memory (LSTM), and convolutional neural networks (CNNs). It simultaneously explores the characteristics of both spatial and temporal features of the received signal for increasing the detection probability of the system. The simulations discuss the advantages of the deep ConvNet-LSTM-based receiver design for downlink NOMA communication systems in terms of bit error rate (BER), achievable rate, and mean squared error (MSE). Furthermore, the robustness of the proposed deep ConvNet-LSTM approach is evaluated for its superiority to other techniques. [ABSTRACT FROM AUTHOR]

The Industrial Internet of Healthcare Things (IIoHT) is the emerging paradigm in digital healthcare. Context-aware healthcare sensors, local intelligent watches, healthcare devices, wireless communication technologies, fog, and cloud computing are all parts of the IIoHT used in healthcare. The ubiquitous healthcare services it provides to its users in practice. However, the current IIoHT healthcare frameworks have security and failure issues in mobile fog and cloud networks where they are spread out. This paper presents the secure, fault-tolerant IIoHT Framework based on digital twin (DT) federated learning-enabled fog-cloud models. The DT is an effective technology that makes virtual copies of servers at different locations. DT integrated with federated learning inside the fog and cloud environments, where the failure of tasks and execution improved for healthcare sensor data. The study aims to reduce processing time and the risk of task failure. The study presents the Secure and Fault-Tolerant Strategies (SFTS)-enabled IIoHT framework that optimizes wearable sensor data and executes it with the minimum offloading and processing delays. Simulation results show that the proposed work minimized the security risk by 40%, failure risk of tasks risk by 50%, and the training and testing time by 39% for sensor data during the execution of mobile fog cloud networks. [ABSTRACT FROM AUTHOR]

• A CNN-LSTM network was utilized for skeleton-based lameness recognition in broilers. • The paper established the first large-scale benchmark for broiler pose estimation, consisting of 9,412 images. • The CNN-LSTM-based model achieved a per-class classification accuracy of 97.5 %. Poultry behavior is a critical indicator of its health and welfare. Lameness is a clinical symptom indicating the existence of health problems in poultry. Therefore, lameness detection in the early stages is vital to broiler producers. In this study, a pose estimation-based model was developed to identify lameness in broilers through analyzing video footages for the first time. A deep convolutional neural network was used to detect and track seven key points on the bodies of walking broilers. Then consecutive extracted key points were fed into Long Short-Term Memory (LSTM) model to classify broilers according to a 6-point assessment method. This paper proposes the first large-scale benchmark for broiler pose estimation, consisting of 9,412 images. In addition, the dataset includes 400 videos (36,120 frames in total) of broilers with different gait score levels. The developed LSTM model achieved an overall classification accuracy of 95%, and the average per class classification accuracy was 97.5%. The obtained results prove that the pose estimation-based model as an automatic and non-invasive tool of lameness assessment can be applied to poultry farms for efficient management. [ABSTRACT FROM AUTHOR]

• A multi-channel acoustic emission real-time efficient signal detection method and acquisition system. • A novel spatio-temporal complementary model (CNN-LSTM) for fusion of time-varying AE signal. • Dual circulation LSTM explains the decay mechanism of AE using temporal sampling points. • Stronger correlation between information from AE attenuation profile and surface quality. Laser shock peening (LSP) is one of the main anti-fatigue technologies in high-end manufacturing industries. However, guaranteeing its quality consistency is difficult due to its process complexity. Acoustic emission (AE) is a promising solution to accurately monitoring the surface quality of LSP, although, the large scale of monitoring data is still challenging for industrial applications. This paper studied the in-situ evaluation of surface residual compressive stress (SRCS) for 7075 aluminum alloy in LSP and the efficient sensing of AE based on deep learning methods. Firstly, a monitoring system of LSP with four types of AE sensors was developed in order to simultaneously acquire both the ultra-high and ultra-low amplitude of shock wave. The performance of those sensors is further quantitatively evaluated via long-short term memory (LSTM). Then, a new spatio-temporal parallel CNN-LSTM model for SRCS classification was proposed and experimentally validated to outperform CNN and LSTM after parameters optimization. Finally, aiming to efficient sensing of AE, the importance of different frames of AE signal was analyzed by means of the proposed dual circulation LSTM (DC-LSTM) model, which can accurately locate the key frame of time series signal. Besides, the effects of different frames of AE on classification performance was discussed. It was found that the broadband AE sensor without attenuator shows the highest classification accuracy, while the fast decay stage of AE signal has more contribution to the SRCS classification. This paper provides guidance for real-time non-destructive evaluation of residual stress via AE in laser manufacturing. [ABSTRACT FROM AUTHOR]

In this paper, we propose a new technique that applies automated image analysis in the area of structural corrosion monitoring and demonstrate improved efficacy compared to existing approaches. Structural corrosion monitoring is the initial step of the risk-based maintenance philosophy and depends on an engineer's assessment regarding the risk of building failure balanced against the fiscal cost of maintenance. This introduces the opportunity for human error which is further complicated when restricted to assessment using drone captured images for those areas not reachable by humans due to many background noises. The importance of this problem has promoted an active research community aiming to support the engineer through the use of artificial intelligence (AI) image analysis for corrosion detection. In this paper, we advance this area of research with the development of a framework, CorrDetector. CorrDetector uses a novel ensemble deep learning approach underpinned by convolutional neural networks (CNNs) for structural identification and corrosion feature extraction. We provide an empirical evaluation using real-world images of a complicated structure (e.g. telecommunication tower) captured by drones, a typical scenario for engineers. Our study demonstrates that the ensemble approach of CorrDetector significantly outperforms the state-of-the-art in terms of classification accuracy. • We presented a novel framework for structural corrosion detection in drone images. • We developed an ensemble deep learning approach to recognise corrosion. • We conducted a comprehensive evaluation using real-world industrial datasets. • We demonstrated the efficacy by comparing with current state-of-the-art models. [ABSTRACT FROM AUTHOR]

• PRNU is a noise pattern extracted from images used to identify the source camera. • Correlation of image and camera reference PRNUs is used for camera identification. • Even slight de-synchronization of PRNU signals can spoil the identification. • This paper presents a novel method to register scaled and rotated PRNU signals. • PRNU registration is obtained using a deep convolutional network. Assessing if an image comes from a specific device is fundamental in many application scenarios. The most promising techniques to solve this problem rely on the Photo Response Non Uniformity (PRNU), a unique trace left during image acquisition. A PRNU fingerprint is computed from several images of a given device, then it is compared with the probe residual noise by means of correlation. However, such a comparison requires that PRNUs are synchronized: even small image transformations can spoil this task. Most of the attempts to solve the registration problem rely on time consuming brute-force search, which is prone to missing detections and false positives. In this paper, the problem is addressed from a computer vision perspective, exploiting recent image registration techniques based on deep learning, and focusing on scaling and rotation transformations. Experiments show that the proposed method is both more accurate and faster than state-of-the-art approaches. [ABSTRACT FROM AUTHOR]

Document forgery is a significant problem for ages due to paper-based documents' pervasive use. Classical destructive approaches for this problem, such as chromatography and electrophoresis, cannot be implemented as they flaw the document under analysis. Hyperspectral imaging - non-destructive approach that assists in finding the unique features of an image under investigation through their unique spectral signatures. It captures multiple narrow-band images at the electromagnetic spectrum, which is difficult through conventional imaging. Deep learning approaches for hyperspectral images have attained state-of-the-art results for solving many complex and challenging problems. Supervised classification of hyperspectral images is a tedious task since obtaining image labels and labeling the training data is a time-consuming and expensive process. In this paper, an unsupervised approach for classification of hyperspectral document images is proposed. To propose an unsupervised deep learning approach for ink mismatch detection in hyperspectral document images using spectral features. CAE-LR approach is proposed that uses Convolutional Autoencoder (CAE) for feature extraction and utilizing them for ink mismatch detection through Logistic Regression (LR). We evaluated the performance of CAE-LR on UWA writing ink hyperspectral images dataset for blue and black inks. Artificially similar color inks of different types (2∼5) were mixed in varying proportions to detect ink mismatch. Additionally, results are compared with three machine learning algorithms with variants of each, CNN, and five state-of-art methods used by the researchers. Experimental results illustrated that the CAE-LR outperforms all the above – mentioned approaches by achieving the state of art results, which depicts the efficacy of unsupervised deep learning approach for ink mismatch detection in hyperspectral document images. [ABSTRACT FROM AUTHOR]

• A unique detection and classification method (DCCNet) for fast diagnosis of COVID-19 patients was presented. • CNN and HOG based feature extractors were used for feature extraction. • YOLOv3 was used to cross check if the input image is a chest X-ray image of a human lung or not. • Combination of CNN and HOG-based features with the SVM classifier achieved 99.97% training accuracy and 99.67% testing accuracy. COVID-19 is now regarded as the most lethal disease caused by the novel coronavirus disease of humans. The COVID-19 pandemic has spread to every country on the planet and has wreaked havoc on these countries by increasing the number of human deaths, and in addition, caused intense hunger, and lowered economic productivity. Due to a lack of sufficient radiologist, a restricted amount of COVID-19 test kits is available in hospitals, and this is also accompanied by a shortage of equipment due to the daily increase in cases, as a result of increase in the number of persons infected with COVID-19. Even for experienced radiologists, examining chest X-rays is a difficult task. Many people have died as a result of inaccurate COVID-19 diagnosis and treatment, as well as ineffective detection measures. This paper, therefore presents a unique detection and classification approach (DCCNet) for quick diagnosis of COVID-19 using chest X-ray images of patients. To achieve quick diagnosis, a convolutional neural network (CNN) and histogram of oriented gradients (HOG) method is proposed in this paper to help medical experts diagnose COVID-19 disease. The diagnostic performance of the hybrid CNN model and HOG-based method was then evaluated using chest X-ray images collected from University of Gondar and online databases. The experiment was performed using Keras (with TensorFlow as a backend) and Python. After the DCCNet model was evaluated, a 99.9% training accuracy and 98.3% test accuracy was achieved, while a 100% training accuracy and 98.5% test accuracy was achieved using HOG. After the evaluation, the hybrid model achieved 99.97% and 99.67% training and testing accuracy for detection and classification of COVID-19 which was better by 1.37% compared to when features were extracted using CNN and 1.17% when HOG was used. The DCCNet achieved a result that outperformed state-of-the-art models by 6.7%. [ABSTRACT FROM AUTHOR]

With the development of the internet and big data, the missing or hidden information identification of text data has become an imperative task. At present, the challenge in the hidden information study is judging whether there is hidden information and where it exists. In this paper, hidden information refers to the words that do not appear in a sentence, however, they have certain correlations with the existing words or sentence and have a great influence on the comprehension of a sentence or part of the text data. This paper focuses on discovering the key and influential hidden information in the text data. A keyword-based hidden information extraction framework is proposed in this paper to search hidden entities, with the assumption that the importance of hidden objects is reflected by the keywords in the text data. A network-based Convolution Neural Network (CNN) model is developed to identify the hidden information related to keywords. The model is based on the results of CNN, and cosine similarity is used to judge whether there is hidden information in the source text data or not. We primarily form the word co-occurrence network of text, select the words with the highest degree as keywords, and generate random walk paths on the network. Besides, we use the random walk path where the last word is the keyword to train CNN. In the experimental section, the proposed model is applied to the dataset in 20Newgroups. The results show that the proposed model can effectively identify the hidden information associated with the keywords in the source text data, and the detection accuracy of keywords can reach 98%–99% achieved by CNN. [ABSTRACT FROM AUTHOR]

In this paper, we tackle the task of similar question retrieval (QR) which is essential for Community Question Answering (cQA) and aims to retrieve historical questions that are semantically equivalent to the new queries. Over time, with the sharp increase of community archives and the accumulation of duplicated questions, the QR problem has become increasingly challenging due to the shortness of the community questions as well as the word mismatch problem as users can formulate the same query using different wording. Although many efforts have been devoted to address this problem, existing methods mostly relied on supervised models which significantly depend on massive training data sets and manual feature engineering. Such methods are chiefly constrained by their specificities that ignore the word order and do not capture enough syntactic and semantic information in questions. In this paper, we rely on Neural Networks (NNs) which use a deep analysis of words and questions to take into consideration the semantics as well as the structure of questions to predict the semantic text similarity. We propose a deep learning approach based on a Siamese architecture with Long Short-Term Memory (LSTM) networks, augmented with an attention mechanism to let the model give different words different attention while modeling questions. We also explore the use of Convolutional Neural Networks (CNN) nested within the Siamese architecture to retrieve relevant questions. Different similarity measures were tested to predict the semantic similarity between the pairs of questions. To evaluate the proposed approach, we conducted experiments on large-scale datasets in English and Arabic. [ABSTRACT FROM AUTHOR]

As the content of the food Review text becomes more and more complex and the types of the food Review text representation become more and more diverse, emotion analysis becomes more and more important in dealing with problems. At present are faced with the problem, this paper puts forward a new type of complementary fusion model, the text pretreatment by Bert training text model, make the content of the food review text more coherent, keep on convolutional neural network logic timing relationships, before and after the statement of text vector of the local feature extraction, using the method of bidirectional long short-term memory network (BLSTM) capture and text context information related to global long-range dependence, using the weighted formula of attention mechanism to improve the comment text statement attention emotional words and then with the weight of characteristic value after conditions with the airport of sorting the output. Through a comparative experiment on the data set of JD food review data set reviews, the results show that BCBLAC, the feature fusion model proposed in this paper, is much more accurate than the existing model in the food Review text language classification. [ABSTRACT FROM AUTHOR]

• The review paper manifest the learning-based schemes which have covered almost the complete domain of image synthesis & analysis. • The paper highlights the critical remarks including comparative analysis for some significant techniques reported so far in the image compression. • The paper pinpoint the problems faced in the successful implementation of deep learning in image compression. Deep learning architectures are now pervasive and filled almost all applications under image processing, computer vision, and biometrics. The attractive property of feature extraction of CNN has solved a lot of conventional image processing problems with much-improved performance & efficiency. The paper aimed to review over a hundred recent state-of-the-art techniques exploiting mostly lossy image compression using deep learning architectures. These deep learning algorithms consists of various architectures like CNN, RNN, GAN, autoencoders and variational autoencoders. We have classified all the algorithms under certain categories for the better and deep understanding. The review is written keeping in mind the contributions of researchers & the challenges faced by them. Various findings for the researchers along with some future directions for a new researcher have been significantly highlighted. Most of the papers reviewed in the compression domain are from the last four years using different methodologies. The review has been summarized by dropping a new outlook for researchers in the realm of image compression. [ABSTRACT FROM AUTHOR]

• 3D-CNNpred is the first 3-dimensional CNN model designed for stock market prediction. • CNNpred successfully combines various sources of information for prediction. • CNNs filters are designed to better handle financial data. • Deep CNN-based framework significantly outperforms shallow ANNs. • CNNpred is profitable in 4 out of 5 tested indices in presence of transaction costs. Feature extraction from financial data is one of the most important problems in market prediction domain for which many approaches have been suggested. Among other modern tools, convolutional neural networks (CNN) have recently been applied for automatic feature selection and market prediction. However, in experiments reported so far, less attention has been paid to the correlation among different markets as a possible source of information for extracting features. In this paper, we suggest a CNN-based framework, that can be applied on a collection of data from a variety of sources, including different markets, in order to extract features for predicting the future of those markets. The suggested framework has been applied for predicting the next day's direction of movement for the indices of S&P 500, NASDAQ, DJI, NYSE, and RUSSELL based on various sets of initial variables. The evaluations show a significant improvement in prediction's performance compared to the state of the art baseline algorithms. [ABSTRACT FROM AUTHOR]

Swarm intelligence algorithms have been widely adopted in solving many highly nonlinear, multimodal problems and have achieved tremendous successes. However, their application on deep neural networks is largely unexplored. On the other hand, deep neural networks, especially convolutional neural network (CNN), have recently achieved breakthroughs in tackling many intractable problems; nevertheless their performance depends heavily on the chosen values of their hyper-parameters, whose fine-tuning is both labor-intensive and time-consuming. In this paper, we propose a novel particle swarm optimization (PSO) variant cPSO-CNN for optimizing the hyper-parameter configuration of architecture-determined CNNs. cPSO-CNN utilizes a confidence function defined by a compound normal distribution to model experts' knowledge on CNN hyper-parameter fine-tunings so as to enhance the canonical PSO's exploration capability. cPSO-CNN also redefines the scalar acceleration coefficients of PSO as vectors to better adapt for the variant ranges of CNN hyper-parameters. Besides, a linear prediction model is adopted for fast ranking the PSO particles to reduce the cost of fitness function evaluation. The experimental results demonstrate that cPSO-CNN performs competitively when compared with several reported algorithms in terms of both CNN hyper-parameter superiority and overall computation cost. [ABSTRACT FROM AUTHOR]