Deep Neural Network-based Model Research Articles

Radar Emitter Signal Intra-Pulse Modulation Open Set Recognition Based on Deep Neural Network

Radar emitter signal intra-pulse modulation recognition is important for modern electronic reconnaissance systems to analyze target radar systems. In the actual environment, the intra-pulse modulations of the sampled radar emitter signals contain not only the known types in the library but also the unknown types. Therefore, the existing recognition methods, which are based on a closed set, cannot recognize the unknown samples. In order to solve this problem, in this paper, we proposed a method for radar emitter signal intra-pulse modulation open set recognition. The proposed method could classify the known modulations and identify the unknown modulation by using an original deep neural network-based recognition model trained on a closed set, estimating the signal-to-noise ratio, and calculating the reconstruction loss by an encoder–decoder model. For a given sample, the original deep neural network-based recognition model will label it as a certain known class temporarily. By estimating the SNR of the sample and calculating the reconstruction loss by inputting the sample to the corresponding encoder–decoder model related to the temporary predicted known class, whether the sample belongs to the predicted temporary known class or the unknown class will be confirmed. Experiments were conducted with five different openness conditions. The experimental results indicate that the proposed method has good performance on radar emitter signal intra-pulse modulation open set recognition.

A Deep Neural Network-Based Assistive Decision Method for Financial Risk Prediction in Carbon Trading Market

The price of carbon emission rights in the market fluctuates greatly due to various factors from economy, finance, and climate. For an enterprise that needs to conduct carbon trading activities, it is extremely necessary to fully grasp the price of carbon trading in the future period. Thus in this paper, a deep neural network-based assistive decision model for financial risk prediction in carbon trading market is proposed for this purpose. Specifically, the dynamic risk spillover effects of domestic and international carbon trading markets are studied, and a frontier time-varying model is utilized to measure the risk spillover effects. Then, the deep neural network is used for quantitative research and to construct an intelligent decision scheme that outputs financial risk prediction results. Four perspectives, energy price, climate environment, carbon market price, and macroeconomics, are selected as the input to analyze the influence factors of carbon emission rights price. Finally, several linear regression models are adopted as the baseline methods for comparison, and experimental results show that the proposed method can achieve better prediction performance compared with baseline methods.

A Local Self-Attention Sentence Model for Answer Selection Task in CQA Systems

Current evidence indicates that the semantic representation of question and answer sentences is better generated by deep neural network-based sentence models than traditional methods in community answer selection tasks. In particular, as a widely recognized language model, the self-attention model computes the similarity between the specific word and the whole sets of words in the same sentence and generates new semantic representation through the similarity-weighted summation of semantic representations of the whole words. However, the self-attention operation entirely considers all the signals with a weighted sum operation, which disperses the distribution of attention, which may result in overlooking the relation of neighboring signals. This issue becomes serious when applying the self-attention model to online community question answering platforms because of the varied length of the user-generated questions and answers. To address this problem, we introduce an attention mechanism enhanced local self-attention (LSA), which restricts the range of original self-attention by a local window mechanism, thereby scaling linearly when increasing the sequence length. Furthermore, we propose stacking multiple LSA layers to model the relationship of multiscale <inline-formula> <tex-math notation="LaTeX">$n$</tex-math> </inline-formula>-gram features. It captures the word-to-word relationship in the first layer and then captures the chunk-to-chunk (such as lexical <inline-formula> <tex-math notation="LaTeX">$n$</tex-math> </inline-formula>-gram phrases) relationship in its deeper layers. We also test the effectiveness of the proposed model by applying the learned representation through the LSA model to a Siamese and a classification network in community question answer selection tasks. Experiments on the public datasets show that the proposed LSA achieves a good performance.

Simulation of neuroplasticity in a CNN-based in-silico model of neurodegeneration of the visual system.

The aim of this work was to enhance the biological feasibility of a deep convolutional neural network-based in-silico model of neurodegeneration of the visual system by equipping it with a mechanism to simulate neuroplasticity. Therefore, deep convolutional networks of multiple sizes were trained for object recognition tasks and progressively lesioned to simulate neurodegeneration of the visual cortex. More specifically, the injured parts of the network remained injured while we investigated how the added retraining steps were able to recover some of the model's object recognition baseline performance. The results showed with retraining, model object recognition abilities are subject to a smoother and more gradual decline with increasing injury levels than without retraining and, therefore, more similar to the longitudinal cognition impairments of patients diagnosed with Alzheimer's disease (AD). Moreover, with retraining, the injured model exhibits internal activation patterns similar to those of the healthy baseline model when compared to the injured model without retraining. Furthermore, we conducted this analysis on a network that had been extensively pruned, resulting in an optimized number of parameters or synapses. Our findings show that this network exhibited remarkably similar capability to recover task performance with decreasingly viable pathways through the network. In conclusion, adding a retraining step to the in-silico setup that simulates neuroplasticity improves the model's biological feasibility considerably and could prove valuable to test different rehabilitation approaches in-silico.

Causal Deep Neural Network-Based Model for First-Line Hypertension Management

ObjectiveTo develop and validate a machine learning model that predicts the most successful antihypertensive treatment for an individual. Patients and MethodsThe causal, deep neural network-based model was trained on data from 16,917 newly diagnosed hypertensive patients attending Mayo Clinic’s primary care practices from January 1, 2005, to December 31, 2021. Eligibility criteria included a diagnosis of primary hypertension, blood pressure and creatinine measurements before antihypertensive treatment, treatment within 9 months of diagnosis, and at least 1 year of follow up. The primary outcome was model performance in predicting the likelihood of a successful antihypertensive treatment 1 year from the start of treatment. Treatment success was defined as achieving blood pressure control with no moderate or severe adverse effects. Model validation and guideline agreement was assessed on 1000 patients. ResultsIn the training set of 16,917 participants (60.8±14.7 years; 8344 [49.3%] women), 33.8% achieved blood pressure control without moderate or severe adverse effects for at least a year with initial treatment. The most common treatment was angiotensin-converting enzyme inhibitor (39.1% average success), and the most successful was angiotensin-converting enzyme inhibitor-thiazide combination (44.4% average success). Our custom-built causal, deep neural network-based model exhibited the highest accuracy in predicting individualized treatment success with a precision of 51.7%, recall of 44.4%, and F1 score of 47.8%. Compared with actual physician practice on the validation set (77.9% agreement), the algorithm aligned with the Eighth Joint National Committee hypertension guidelines 95.7% of the time. ConclusionA machine learning algorithm can accurately predict the likelihood of antihypertensive treatment success and help personalize hypertension management.

SAAC-Net: deep neural network-based model for atmospheric correction in remote sensing

ABSTRACT Atmospheric correction eliminates corruption in reflectance captured by satellite images due to atmospheric elements like gases, aerosols, and water vapours. Existing physics-based approaches employ radiative transfer models constructed using lookup tables computed for different atmospheric parameters. However, these approaches are computationally expensive and rely on estimates of parameters that are difficult to sense accurately. This paper proposes a deep learning model as an alternative to physics-based approaches. We present an end-to-end deep neural network trained on seasonally and spatially rich Landsat 8 satellite images without explicit atmospheric parameterization along with our analysis and its validation. We validate the model’s effectiveness vis-a-vis Landsat 8’s Land Surface Reflectance Code – LaSRC results in , , and correlation coefficient . For ground measurements by RadCalNet, the proposed model has an , , and . The results show that the model accurately predicts surface reflectance and correlates highly with reference data.

Invariance embedded physics-infused deep neural network-based sub-grid scale models for turbulent flows

In this paper, we present two novel physics-infused neural network (NN) architectures that satisfy various invariance conditions for constructing efficient and robust Deep Learning (DL)-based sub-grid scale (SGS) turbulence models for use in Large Eddy Simulation (LES) procedures widely used in various fluid engineering applications. The first architecture, called tensor basis neural networks (TBNN), recently proposed in the context of Reynolds-averaged Navier–Stokes (RANS) modeling, and introduced herein for SGS modeling of wall-bounded turbulence, embeds the analytical expansion of the SGS turbulence stresses into integrity basis tensors composed of the symmetric and anti-symmetric parts of the resolved velocity gradient tensor, thus incorporating the Galilean, rotational and reflectional invariances. In our second approach, a relatively simple yet powerful architecture, called Galilean Invariance embedded Neural Network (GINN) incorporates the Galilean invariance, and takes in as inputs the independent components of the integrity basis tensors in addition to the invariant inputs in a single input layer. Explicit filtering of data from direct simulations of the canonical channel flow at two friction Reynolds numbers Reτ≈395 and 590 provided accurate data for training and testing these models. Both sets of models are used to predict the SGS stresses for feature datasets generated with different filter widths, and at different Reynolds numbers. The GINN model yields less prediction error on test datasets (mean squared error ∼0.4) compared to the TBNN model (mean squared error ∼0.5). Upon comparison, it is revealed that GINN has better feature extraction capacity owing to its ability to establish relations between and extract information from cross-components of the integrity basis tensors and the SGS stresses. Based on their predictive performance, both models proposed herein have shown great promise for use in a posteriori actual LESs. The present work illustrates the significance of physics infusion as well as invariance embedding into NN architecture for constructing efficient and robust DL-based SGS turbulent models in achieving superior predictive efficacy.

Meaning-Sensitive Text Data Augmentation with Intelligent Masking

With the recent popularity of applying large-scale deep neural network-based models for natural language processing (NLP), attention to develop methods for text data augmentation is at its peak, since the limited size of training data tends to significantly affect the accuracy of these models. To this end, we propose a novel text data augmentation technique called Intelligent Masking with Optimal Substitutions Text Data Augmentation (IMOSA). IMOSA, developed for labelled sentences, can identify the most favourable sentences and locate the appropriate word combinations in a particular sentence to replace and generate synthetic sentences with a meaning closer to the original sentence, while also significantly increasing the diversity of the dataset. We demonstrate that the proposed technique notably improves the performance of classifiers based on attention-based transformer models through the extensive experiments for five different text classification tasks which are performed under the low data regime in a context-aware NLP setting. The analysis clearly shows that IMOSA effectively generates more sentences using favourable original examples and completely ignores undesirable examples. Furthermore, the experiments carried out confirm IMOSA’s ability to add diversity to the augmented dataset using multiple distinct masking patterns against the same original sentence, which remarkably adds variety to the training dataset. IMOSA consistently outperforms the two key masked language model-based text data augmentation techniques, and demonstrates a robust performance against the critical challenging NLP tasks.

Identification of glaucoma in fundus images utilizing gray wolf optimization with deep convolutional neural network-based resnet50 model

Identification of glaucoma in fundus images utilizing gray wolf optimization with deep convolutional neural network-based resnet50 model

Firms' multi-sided platform construction efforts and ESG performance: An information processing theory perspective

This research explores the impact of business-to-business (B2B) firms' efforts to construct technology-enabled multi-sided platforms (MSP efforts) on their ESG performance through an information processing theory (IPT) lens. Integrating the literature on digital transformation and MSPs, we delineate firms' MSP efforts along with three dimensions (digital operation efforts, technology sharing efforts, and flat management efforts) and posit that each addresses specific challenges associated with the environmental, social, and governance components of ESG performance. Using a sample of 213 publicly listed Chinese firms, encompassing 1527 firm-year observations, we measure the variation in the three dimensions of MSP efforts through a deep neural network-based latent Dirichlet allocation topic model to analyze their annual reports. We find that digital operation efforts enhance environmental, social, and governance performance; technology sharing efforts primarily improve environmental and social performance; and flat management efforts boost only governance performance. In addition, these effects are more pronounced for large firms than for small firms. Two robustness checks, employing patent applications and granted patents as alternative indicators of firms' MSP efforts, further corroborate these findings. This research augments both the MSP and ESG literature, offering actionable insights for firms, governments, and non-governmental organizations.

Anatomically Guided Cross-Domain Repair and Screening for Ultrasound Fetal Biometry.

Ultrasound based estimation of fetal biometry is extensively used to diagnose prenatal abnormalities and to monitor fetal growth, for which accurate segmentation of the fetal anatomy is a crucial prerequisite. Although deep neural network-based models have achieved encouraging results on this task, inevitable distribution shifts in ultrasound images can still result in severe performance drop in real world deployment scenarios. In this article, we propose a complete ultrasound fetal examination system to deal with this troublesome problem by repairing and screening the anatomically implausible results. Our system consists of three main components: A routine segmentation network, a fetal anatomical key points guided repair network, and a shape-coding based selective screener. Guided by the anatomical key points, our repair network has stronger cross-domain repair capabilities, which can substantially improve the outputs of the segmentation network. By quantifying the distance between an arbitrary segmentation mask to its corresponding anatomical shape class, the proposed shape-coding based selective screener can then effectively reject the entire implausible results that cannot be fully repaired. Extensive experiments demonstrate that our proposed framework has strong anatomical guarantee and outperforms other methods in three different cross-domain scenarios.

Integrated Disaster Risk Management for Flood Detection on Remote Sensing Images using Deep Learning techniques

&lt;p&gt;Floods are one of the leading causes of damage, prompting mortality and substantial destruction to the structure and total economy of the affected nations. Remote sensing, satellite imagery, global positioning system, and geographic information system (GIS) are widely employed for flood identification to examine flood-related losses. Recently, accurate and automated flood detection models using remote sensing images have become effective for flood disaster management, risk manager, infrastructure planning, disaster rescue management, etc. Computer vision and deep learning (DL) models provide prompt and rapid flood detection in remote sensing images. In this aspect, this paper presents a multiverse optimization with a deep transfer learning-enabled flood detection (MVODTL-FD) technique for disaster risk management. In the proposed MVODTL-FD technique, remote sensing images are investigated for the effectual detection of floods. To accomplish this, the presented MVODTL-FD technique applies a guided normal filter (GNF) based image preprocessing approach to eliminate the noise. In addition, the proposed MVODTL-FD technique uses a deep convolutional neural network-based Squeeze Net model for feature extraction, and the hyperparameter process is performed using the MVO algorithm. At last, the flood detection process is performed using support vector machine (SVM) classification. For establishing the improved version of the MVODTL-FD method, a wide-ranging experimental analysis is performed. The MVODTL-FD model is rated higher in the comparative analysis than other DL models.&lt;/p&gt;&#x0D;

Identification of Fake News Using Deep Neural Network-Based Hybrid Model

Identification of Fake News Using Deep Neural Network-Based Hybrid Model

Deep Learning Based Automated Image Captioning System

With the rise of deep learning in recent years, the merging of computer vision and natural language processing has inspired a huge interest. Image captioning is an unique deep learning application that has advanced rapidly in recent years. The ability to train a computer to properly describe the content of an image or an environment like a person has huge implications in computer vision, economics, and a variety of other fields. Across the year, this has been a difficult issue in the field of artificial intelligence, and many researchers have made tremendous progress. In this paper, we describe various deep neural network-based picture caption generation models, including RNN-based decoding and CNN-based accessing and Reinforcement-based framework. We also created captions for some of the images and compared the various feature extraction and encoder models to see which one provides the best accuracy and produces the desired outcome. Extensive experiments on the dataset reveal that the proposed framework outperforms the existing encoder based approach.

Classification and concentration estimation of CO and NO2 mixtures under humidity using neural network-assisted pattern recognition analysis

This study addresses the concerns regarding the cross-sensitivity of metal oxide sensors by building an array of sensors and subsequently utilizing machine earning techniques to analyze the data from the sensor arrays. Sensors were built using In2O3, Au-ZnO, Au-SnO2, and Pt-SnO2 and they were operated simultaneously in the presence of 25 different concentrations of nitrogen dioxide (NO2), carbon monoxide (CO), and their mixtures. To investigate the effects of humidity, experiments were conducted to detect 13 distinct CO and NO2 gas combinations in atmospheres with 40% and 90% relative humidity. Principal component analysis was performed for the normalized resistance variation collected for a particular gas atmosphere over a certain period, and the results were used to train deep neural network-based models. The dynamic curves produced by the sensor array were treated as pixelated images and a convolutional neural network was adopted for classification. An accuracy of 100% was achieved using both models during cross-validation and testing. The results indicate that this novel approach can eliminate the time-consuming feature extraction process.

Learning knowledge graph embedding with multi-granularity relational augmentation network

Knowledge graph embedding (KGE) aims to complete link prediction tasks effectively by learning the representation of entity and relation. Recently, deep neural networks have achieved prominent results for learning KGE. However, knowledge graphs typically contain massive information about entities and relations, and existing deep neural network-based KGE models exploit semantic information from simple explicit feature concatenation and reshaping without considering bidirectional implicit interactions, which cannot accentuate relevant predictive information. Additionally, these models simply extract nonlinear features from embedded matrices of relations, resulting in neglecting the distinctive semantic connotation of heterogeneous relations. To overcome these two challenges, a multi-granularity relational augmentation network (MRAN) to learn KGE is proposed. Specifically, multi-granularity implicit interaction embeddings are generated to facilitate relevant predictive information selection, which model bidirectional implicit interactions between entities and relations in Euclidean space, complex space, and quaternion space. Moreover, relational augmentation convolution module utilizes relation-aware filters at the convolutional layer to preserve the distinctive relational properties for dealing with heterogeneous relations. Subsequently, the augmentation mechanism is injected to focus on informative features and restrain useless features of neural networks, which boosts the quality of KGE. Experimental results demonstrate the outstanding link prediction results of MRAN on five benchmark datasets, revealing the benefits of considering bidirectional implicit interactions and heterogeneous relations. Our model substantially improves efficiency while achieving excellent performance, which is superior to state-of-the-art baseline models in 2 out of 4 metrics on WN18 and in 3 out of 4 metrics on FB15k-237 and YAGO3-10 datasets.

Deep Neural Network-Based Ensemble Model for Eye Diseases Detection and Classification

Fundus images are the principal tool for observing and recognizing a wide range of ophthalmological abnormalities. The automatic and robust methods based on color fundus images are urgently needed since few symptoms are observable in the early stages of the disease. Experts must manually evaluate images to detect diseases for screening procedures to be effective. Due to the complexity of the screening procedure and the shortage of experienced personnel, developing successful screening-based treatments is costly. Although existing automated approaches strive to address these issues, they cannot handle a wide range of diseases and real-world circumstances. We design an automated deep learning-based ensemble method to detect and classify eye diseases from fundus images to address the abovementioned problems. A deep CNN-based model is proposed in the ensemble method that incorporates a mix of 20 layers, including the activation, optimization, and loss functions. The contrast-limited adaptive histogram equalization (CLAHE) and Gaussian filter are utilized in the pre-processing step to get more explicit images and eliminate noise. To avoid overfitting in the training phase, augmentation techniques are applied. Three pre-trained CNN models, including VGG16, DenseNet201, and ResNet50, are employed to compare and assess the efficiency of the proposed CNN model. Experimental results demonstrate that the ensemble approach outperforms recent approaches, which is comparatively state-of-art in the ODIR publicly available dataset.

DeepAlgPro: an interpretable deep neural network model for predicting allergenic proteins.

Allergies have become an emerging public health problem worldwide. The most effective way to prevent allergies is to find the causative allergen at the source and avoid re-exposure. However, most of the current computational methods used to identify allergens were based on homology or conventional machine learning methods, which were inefficient and still had room to be improved for the detection of allergens with low homology. In addition, few methods based on deep learning were reported, although deep learning has been successfully applied to several tasks in protein sequence analysis. In the present work, a deep neural network-based model, called DeepAlgPro, was proposed to identify allergens. We showed its great accuracy and applicability to large-scale forecasts by comparing it to other available tools. Additionally, we used ablation experiments to demonstrate the critical importance of the convolutional module in our model. Moreover, further analyses showed that epitope features contributed to model decision-making, thus improving the model's interpretability. Finally, we found that DeepAlgPro was capable of detecting potential new allergens. Overall, DeepAlgPro can serve as powerful software for identifying allergens.

Single image dehazing using extended local dark channel prior

Image dehazing is an important technique aimed at eliminating the haze in the atmosphere to enhance the image's visual quality. There are many applications where it has been used as a prepossessing step, such as in event detection. In recent years, the dark channel prior methodology has been recognised as an effective approach for eliminating haze from hazy images. However, the main drawback of the existing dark channel prior methodology is that it only considers a single colour channel of the RGB image with pixels having minimum intensity values. This non-uniform selection of the dark channel from a single channel eradicates the effect of the transmission across the different channels of the hazy image. Hence, the haze cannot be removed to a great extent using the existing method. Thus, to address the problem of non-uniform estimation of the dark channel by the existing dark channel prior method, we propose an approach where the dark channel will be computed from all three channels of an image by selecting the minimum intensity. The main advantage of using the proposed prior-based methodology for image dehazing over deep neural network-based models such as CNN or GANs is that training deep models requires a large amount of training data, thus resulting in a longer training time. Experimental outcomes exhibit that the proposed technique outperforms state-of-the-art methods on synthetic datasets as well as real-world hazy images. The findings demonstrate that the proposed technique obtains better accuracy as compared to the state-of-the-art methods and recent deep learning-based models over the D-HAZY, I-HAZE, O-HAZE and Middlebury databases.

Deep neural network-based analysis of the impact of ambidextrous innovation and social networks on firm performance

The motivation for analyzing the impact of deep neural networks on enterprise performance is mainly due to the continuous deepening of enterprise information construction, shifting from traditional paper-based data acquisition methods to electronic data management. The data generated by the sales, production, logistics and other links of enterprises is also becoming increasingly large. How to scientifically and effectively process these massive amounts of data and extract valuable information has become an important issue that enterprises need to solve. The continuous and stable growth of China's economy has promoted the development and growth of enterprises, however, it has also made enterprises face a more complex competitive environment. The question of how to improve the performance of enterprises to enhance their competitiveness in the market has become a major issue to be addressed in the face of fierce competition and to ensure the long-term development of enterprises. In this paper, based on the research of firm performance evaluation, deep neural network is introduced to analyse the influence of ambidextrous innovation and social network on firm performance, and the theories of social network, ambidextrous innovation and deep neural network are sorted out and analysed, and a deep neural network-based firm performance evaluation model is established, and finally the sample data is obtained using crawler technology, and then the response values are analysed. Innovation and the improvement of the mean value of social networks are helpful to firm performance.