Deep Neural Network Techniques Research Articles

Query-driven video event processing for the internet of multimedia things

Advances in Deep Neural Network (DNN) techniques have revolutionized video analytics and unlocked the potential for querying and mining video event patterns. This paper details GNOSIS, an event processing platform to perform near-real-time video event detection in a distributed setting. GNOSIS follows a serverless approach where its component acts as independent microservices and can be deployed at multiple nodes. GNOSIS uses a declarative query-driven approach where users can write customize queries for spatiotemporal video event reasoning. The system converts the incoming video streams into a continuous evolving graph stream using machine learning (ML) and DNN models pipeline and applies graph matching for video event pattern detection. GNOSIS can perform both stateful and stateless video event matching. To improve Quality of Service (QoS), recent work in GNOSIS incorporates optimization techniques like adaptive scheduling, energy efficiency, and content-driven windows. This paper demonstrates the Occupational Health and Safety query use cases to show the GNOSIS efficacy.

Breast Cancer Detection using Deep Learning Techniques

Breast cancer is the most common form of cancer among women and the second most common cancer in the world (an estimated 1 152 161 new cases per year), trailing only lung cancer .The current approach to this disease involves early detection and treatment. This approach in the United States yields an 85% 10-year survival rate. Survival is directly related to stage at diagnosis, as can be seen by a 98% 10- year survival rate for patients with stages 0 and I disease compared with a 65% 10-year survival rate for patients with stage III disease. To improve survival in this disease, more patients need to be identified at an early stage.Therefore, we evaluated existing and emerging technologies used for breast cancer screening and detection to identify areas for potential improvement. The main criteria for a good screening test are accuracy, high sensitivity, ease of use, acceptability to the population being screened (with regard to discomfort and time), and low cost. We can begins by describing commonly used breast cancer detection techniques and then delves into emerging modalities. Several studies addressing breast cancer using Deep learning techniques. Many claim that their algorithms are faster, easier, or more accurate than others . This system is based on thermal image processing and Deep learning algorithms that aim to construct a system to accurately differentiate between benign and malignant breast tumors. The aim of this was to optimize the learning algorithm. In this system , we applied the deep neural network technique to select the best features and perfect parameter values of the deep machine learning. The present study proves that deep neural network can automatically find the best model by combining feature preprocessing methods and classification algorithms.

Novel deep neural network technique for detecting environmental effect of COVID-19

ABSTRACT The studies claim that COVID-19 has positive impacts on the environment because it minimizes air pollution, water pollution, and noise pollution due to lockdown. On the contrary, COVID-19 is harming the environment due to increased medical wastage. COVID-19 has been declared a pandemic by the World Health Organization (WHO). Due to the exponential growth of COVID-19 cases people using large quantities of medical accessories to shield themselves from coronavirus, a large amount of medical wastage is produced per day. This medical wastage is a major concern for the expert because this medical waste is not adequately handled. The early detection of the COVID patient is the only solution to control this coronavirus. Several COVID detection models have been proposed in the last few months. Most of the existing models have a high false-positive rate where COVID patients are classified as healthy. To address this problem, this paper explores the positive and negative environmental consequences of COVID-19 and suggests a novel method based on artificial intelligence (AI) to identify COVID-19 disease. A comparative analysis of different previously trained models such as Visual Geometry Group Network (VGGNet-19), Residual Network (ResNet50), and Inception ResNet V2 is presented in this paper. Experimental results show that Inception_ResNet_V2 is a better choice for COVID detection. It has a minimal false-positive rate and offers 99.26% and 94% higher training and test accuracy compared to VGGNet and ResNet, respectively.

The use of deep neural networks for developing generic pavement rutting predictive models

ABSTRACT Rutting prediction models are essential elements of efficient pavement management systems. Accuracy of commonly used predictive models necessitates knowledge of the input parameters that they incorporate and local calibration of the model coefficients. This study aims at developing a rutting prediction model that incorporates a limited number of inputs, yet is able to accommodate, with sufficient generalisation abilities, to the data scarcity and resource limitations in developing countries. The prediction model is developed by employing deep neural network techniques (DNN) on data extracted from the Long-Term Pavement Performance (LTPP) database. The predictive capability of the DNN model is compared to those of the state-of-the-practice and a multivariate linear regression model fitted using the same dataset. It is found that the DNN rutting prediction model features enhanced predictive power compared to commonly used models in the literature. In addition to predicting pavement rutting, the developed model is further utilised to assess and rank the relative impact of the different model inputs on rutting. The sensitivity analysis results confirm the high influence of traffic and climatic conditions. Moreover, generic family rutting predictive curves corresponding to specific traffic, climate, and performance combinations are developed to render rutting predictions available to all road agencies.

Deep Learning Assisted Neonatal Cry Classification via Support Vector Machine Models.

Neonatal infants communicate with us through cries. The infant cry signals have distinct patterns depending on the purpose of the cries. Preprocessing, feature extraction, and feature selection need expert attention and take much effort in audio signals in recent days. In deep learning techniques, it automatically extracts and selects the most important features. For this, it requires an enormous amount of data for effective classification. This work mainly discriminates the neonatal cries into pain, hunger, and sleepiness. The neonatal cry auditory signals are transformed into a spectrogram image by utilizing the short-time Fourier transform (STFT) technique. The deep convolutional neural network (DCNN) technique takes the spectrogram images for input. The features are obtained from the convolutional neural network and are passed to the support vector machine (SVM) classifier. Machine learning technique classifies neonatal cries. This work combines the advantages of machine learning and deep learning techniques to get the best results even with a moderate number of data samples. The experimental result shows that CNN-based feature extraction and SVM classifier provides promising results. While comparing the SVM-based kernel techniques, namely radial basis function (RBF), linear and polynomial, it is found that SVM-RBF provides the highest accuracy of kernel-based infant cry classification system provides 88.89% accuracy.

A Product Pose Tracking Paradigm Based on Deep Points Detection

The paper at hand presents a novel and versatile method for tracking the pose of varying products during their manufacturing procedure. By using modern Deep Neural Network techniques based on Attention models, the most representative points to track an object can be automatically identified using its drawing. Then, during manufacturing, the body of the product is processed with Aluminum Oxide on those points, which is unobtrusive in the visible spectrum, but easily distinguishable from infrared cameras. Our proposal allows for the inclusion of Artificial Intelligence in Computer-Aided Manufacturing to assist the autonomous control of robotic handlers.

Automatic detection of burial mounds (kurgans) in the Altai Mountains

The Altai Mountains are one of the most impressive and valuable archaeological areas in the world. Kurgans (burial mounds) of ancient civilizations, which are scattered across the vast Altai area, are an exceptionally valuable source of information for archaeology. These precious archaeological resources, which sometimes have been preserved intact in the permafrost underground for over two millennia, are now under various threats, such as natural disasters, farmland expansion, touristic development, and most notably global warming. A detailed map or inventory of the mounds is essential but is still not available. In this study, we test the deep convolutional neural network (CNN) technique for automatic detection of stone mounds from high-resolution satellite images in four regions in the Altai Mountains. We propose three improvement techniques to increase the performance of off-the-shelf object detection methods that are originally proposed for daily-life objects. Our results demonstrate that it is feasible to apply CNN to detect stone mounds, and the detection results are good enough to capture their spatial distribution. CNN-based object detection can largely narrow down the search area for archaeologists in yet un-surveyed regions, and is therefore useful for preparing field survey campaigns and directing archaeological fieldwork. We also applied the method to an un-surveyed Altai Mountain area and successfully discovered stone mounds that are yet undocumented. Our method can potentially be applied to construct an inventory for all stone mounds present in the whole Altai Mountain region.

Modeling Hot-Mix asphalt dynamic modulus using deep residual neural Networks: Parametric and sensitivity analysis study

The dynamic modulus (E*) of hot-mix asphalt mixtures is one of the most tedious and time-consuming laboratory testing material properties. It requires costly, advanced equipment and skills that are not easily accessible in the majority of laboratories yet. Thus, many studies have been dedicated to developing E* predictive models. Unfortunately, it is a complex task due to the many input variables and their non-linear effect on the E*. This study applies a deep residual neural networks (DRNNs) technique for the first time to the problem to enhance the E* prediction capabilities. The proposed DRNNs architecture utilizes residual connections (i.e., shortcuts) that bypass some layers in the deep network structure in order to alleviate the problem of training with high accuracy. An intensive laboratory database is employed in the DRNNs model development considering all influential input parameters such as; mixture gradation, volumetric properties, binder characteristics, and testing conditions parameters. Moreover, a brute force enumeration is integrated in the model to reduce the number of needed input variables and identify the best combinations of them. Then, the proposed DRNNs performance, with the best combination of inputs, is evaluated using representative performance indicators and compared with the well-known E* predictive models, namely; Witczak 1-37A, Witczak 1-40D, and Hirsch models. Finally, a variance-based global sensitivity (VB-GS) analysis is conducted with the Monte Carlo simulation aid to highlight each input variable effect on the E* magnitude in real practice while removing the potential distortion of results due to the input variables correlations. Performance evaluation indicators reveal that the DRNNs model outperforms other E* prediction ones. Furthermore, VB-GS analysis shows that, among all feasible inputs, binder stiffness characteristics and testing temperature are the most significant ones.

Automated accurate emotion recognition system using rhythm-specific deep convolutional neural network technique with multi-channel EEG signals

Emotion is interpreted as a psycho-physiological process, and it is associated with personality, behavior, motivation, and character of a person. The objective of affective computing is to recognize different types of emotions for human-computer interaction (HCI) applications. The spatiotemporal brain electrical activity is measured using multi-channel electroencephalogram (EEG) signals. Automated emotion recognition using multi-channel EEG signals is an exciting research topic in cognitive neuroscience and affective computing. This paper proposes the rhythm-specific multi-channel convolutional neural network (CNN) based approach for automated emotion recognition using multi-channel EEG signals. The delta (δ), theta (θ), alpha (α), beta (β), and gamma (γ) rhythms of EEG signal for each channel are evaluated using band-pass filters. The EEG rhythms from the selected channels coupled with deep CNN are used for emotion classification tasks such as low-valence (LV) vs. high valence (HV), low-arousal (LA) vs. high-arousal (HA), and low-dominance (LD) vs. high dominance (HD) respectively. The deep CNN architecture considered in the proposed work has eight convolutions, three average pooling, four batch-normalization, three spatial drop-outs, two drop-outs, one global average pooling and, three dense layers. We have validated our developed model using three publicly available databases: DEAP, DREAMER, and DASPS. The results reveal that the proposed multivariate deep CNN approach coupled with β-rhythm has obtained the accuracy values of 98.91%, 98.45%, and 98.69% for LV vs. HV, LA vs. HA, and LD vs. HD emotion classification strategies, respectively using DEAP database with 10-fold cross-validation (CV) scheme. Similarly, the accuracy values of 98.56%, 98.82%, and 98.99% are obtained for LV vs. HV, LA vs. HA, and LD vs. HD classification schemes, respectively, using deep CNN and θ-rhythm. The proposed multi-channel rhythm-specific deep CNN classification model has obtained the average accuracy value of 57.14% using α-rhythm and trial-specific CV using DASPS database. Moreover, for 8-quadrant based emotion classification strategy, the deep CNN based classifier has obtained an overall accuracy value of 24.37% using γ-rhythms of multi-channel EEG signals. Our developed deep CNN model can be used for real-time automated emotion recognition applications.

A system for automatic rice disease detection from rice paddy images serviced via a Chatbot

A LINE Bot System to diagnose rice diseases from actual paddy field images was developed and presented in this paper. It was easy-to-use and automatic system designed to help rice farmers improve the rice yield and quality. The targeted images were taken from the actual paddy environment without special sample preparation. We used a deep learning neural networks technique to detect rice diseases from the images. We developed an object detection model training and refinement process to improve the performance of our previous research on rice leave diseases detection. The process was based on analyzing the model’s predictive results and could be repeatedly used to improve the quality of the database in the next training of the model. The deployment model for our LINE Bot system was created from the selected best performance technique in our previous paper, YOLOv3, trained by refined training data set. The performance of the deployment model was measured on 5 target classes and found that the Average True Positive Point improved from 91.1% in the previous paper to 95.6% in this study. Therefore, we used this deployment model for Rice Disease LINE Bot system. Our system worked automatically real-time to suggest primary diagnosis results to the users in the LINE group, which included rice farmers and rice disease specialists. They could communicate freely via chat. In the real LINE Bot deployment, the model’s performance was measured by our own defined measurement Average True Positive Point and was found to be an average of 78.86%. The system was fast and took only 2–3 s for detection process in our system server.

A Deep Learning Neural Network Techniques in Visualization, Imaging Data Acquisition And Diagnosis for Covid-19

The corona virus disease pandemic of 2019 (COVID-19) is sweeping the globe. Medical imaging, such as X-ray and computed tomography (CT), is critical in the global fight against COVID-19, and recently evolving artificial intelligence (AI) technologies are enhancing the capacity of imaging tools and assisting medical specialists. For example, image acquisition driven by Deep Learning Architecture may help optimise the scanning process and reshape the workflow with minimal patient intervention, ensuring the best security for imaging technicians. Furthermore, computer-aided platforms assist radiologists in making clinical decisions, such as disease identification, surveillance, and prognosis. In this workflow, we cover the full range of COVID-19-related medical imaging and analysis techniques, including image processing, segmentation, diagnosis, and follow-up. Traditional methods are used to interpret the evaluation, and various output metrics are collected.

Detection of pneumonia infection in lungs from chest X-ray images using deep convolutional neural network and content-based image retrieval techniques.

In this research, A Deep Convolutional Neural Network was proposed to detect Pneumonia infection in the lung using Chest X-ray images. The proposed Deep CNN models were trained with a Pneumonia Chest X-ray Dataset containing 12,000 images of infected and not infected chest X-ray images. The dataset was preprocessed and developed from the Chest X-ray8 dataset. The Content-based image retrieval technique was used to annotate the images in the dataset using Metadata and further contents. The data augmentation techniques were used to increase the number of images in each of class. The basic manipulation techniques and Deep Convolutional Generative Adversarial Network (DCGAN) were used to create the augmented images. The VGG19 network was used to develop the proposed Deep CNN model. The classification accuracy of the proposed Deep CNN model was 99.34 percent in the unseen chest X-ray images. The performance of the proposed deep CNN was compared with state-of-the-art transfer learning techniques such as AlexNet, VGG16Net and InceptionNet. The comparison results show that the classification performance of the proposed Deep CNN model was greater than the other techniques.

Spatio-temporal distribution of Gymnocypris przewalskii during migration with UAV-based photogrammetry and deep neural network

The naked carp (Gymnocypris przewalskii) plays a central role in the ecosystem of the Qinghai Lake, the largest saline-alkaline lake in China. The adult naked carp migrates in large groups with high population density annually from the Qinghai Lake to the upstream freshwater rivers to spawn. Nevertheless, the responsiveness of the fish to local abiotic cues in the form of distribution patterns during migration across the riverine-lacustrine network of the Qinghai Lake region remains unknown. This knowledge gap has reduced efficiency in fish conservation and management efforts in the region. To address this issue, we carried out two field surveys from June to August, 2018, with the aid of unmanned aerial vehicles to a 200-m long back channel characterizing diverse morphological and hydraulic features on the migration route. Combined structure from motion photogrammetry and deep neural network techniques were used to establish a new workflow for detecting and extracting the profiles of fish individuals in large schools. The spatio-temporal distribution pattern of the fish demonstrated that the naked carp was attracted by hydraulic environments with high flow velocity or deep-water during migration. The diurnal variation of temperature and light could alter the preference for hydraulic environments of the fish. Our results highlight the crucial role of the interactions between river morphology and hydraulics, water temperature and light on the migration behaviours of the naked carp.

Artificial Bee Colony and Deep Neural Network-Based Diagnostic Model for Improving the Prediction Accuracy of Diabetes

A large number of machine learning approaches are implemented in healthcare field for effective diagnosis and prediction of different diseases. The aim of these machine learning approaches is to build automated diagnostic tool for helping the physician as well as monitor the health status of patients. These diagnostic tools are widely adopted in intensive care unit for life expectancy of patients. In this study, an effort is made to design an automated diagnostic model for the diagnosis and prediction of diabetes patients. The proposed diagnostic model is designed using artificial bee colony (ABC) algorithm and deep neural network (DNN) technique, called ABC-DNN-based diagnostic model. The ABC algorithm is applied to determine the relevant features for diabetes prediction and diagnosis while DNN technique is adopted for the prediction and diagnosis of diabetes affected patients. The performance of proposed diagnostic model is tested over Pima Indian Diabetes dataset and evaluated using accuracy, sensitivity, specificity, F-measure, Kappa, and area under curve (AUC) parameters. Further, 10-fold and 50-50% training-testing method are considered to assess the performance of proposed diagnostic model. The experimental results of proposed ABC-DNN model is compared with DNN technique and several existing diabetes studies. It is observed that proposed ABC-DNN model achieves 94.74% accuracy rate using 10-fold method.

Deep neural network and model-based clustering technique for forensic electronic mail author attribution

Electronic mail is the primary source of different cyber scams. Identifying the author of electronic mail is essential. It forms significant documentary evidence in the field of digital forensics. This paper presents a model for email author identification (or) attribution by utilizing deep neural networks and model-based clustering techniques. It is perceived that stylometry features in the authorship identification have gained a lot of importance as it enhances the author attribution task's accuracy. The experiments were performed on a publicly available benchmark Enron dataset, considering many authors. The proposed model achieves an accuracy of 94% on five authors, 90% on ten authors, 86% on 25 authors and 75% on the entire dataset for the Deep Neural Network technique, which is a good measure of accuracy on a highly imbalanced data. The second cluster-based technique yielded an excellent 86% accuracy on the entire dataset, considering the authors' number based on their contribution to the aggregate data.

Novel Deep Neural Network Model for Handwritten Digit Classification and Recognition

Deep neural network is a technique of deep learning, where deep neural network model have multiple hidden layers with input and output layer, but artificial neural network have single hidden layer between input and output layer. The use of multiple hidden layers in deep neural network is to improve the performance of model and achieving the higher accuracy compare to machine learning models and their accuracy. The field of pattern recognition is mostly used by the researchers for their research work. There are lots of pattern are available in the field of pattern recognition like: handwritten digits, characters, images, faces, sound, speech etc. In this paper we have concentrated on handwritten digits classification and recognition. For handwritten digit datasets, we have used commonly known Arkiv Digital Sweden (ARDIS) [1] dataset and United State postal service (USPS) [7] dataset. ARDIS dataset is a collection of 7600 samples, where 6600 used as training samples and 1000 used as testing samples. USPS dataset is a collection of 10000 image samples where 7291 samples are used as training sample and 2007 samples are used as testing samples. In this paper we have implemented the proposed deep neural network technique for the classification and recognition of the ARDIS and USPS dataset. The proposed model has collection of 6 layers with relu and softmax activation function. After implementing model, 98.70% testing and 99.76% training accuracy for ARDIS samples achieved, which is higher than previous research accuracy. Also 98.22% training and 93.01%testing accuracy with USPS samples dataset has been achieved. The results represents the performance of deep neural networks have been outstanding compare to other previous techniques.

Image-based recognition of green perilla leaves using a deep neural network for robotic harvest support

This paper describes a method of recognizing green perilla leaves using a deep neural network for harvest support in greenhouse horticulture. We are developing a robot for harvest support, which automates the selection and bundling process. In order to manipulate green perilla leaves correctly, the robot needs to precisely estimate their geometrical parameters such as width, height, and orientation. It also needs to detect leaves with anomalies. Therefore, we develop an image-based leaf recognition method, adopting deep neural network (DNN) techniques. To reduce computation time, we design a network for executing multiple tasks simultaneously, namely, segmentation and classification. We also developed an annotated dataset using conventional image processing techniques. Experimental results show the efficiency and effectiveness of the proposed method.

A Novel Approach for Sentiment Analysis Using Deep Recurrent Networks and Sequence Modeling

Background: Due to the increasing growth of social websites, a lot of user-generated data is available these days in the form of customer reviews, opinions, and comments. Objective: Sentiment analysis includes analyzing the user reviews and finding the overall opinions from the reviews in terms of positive, negative and neutral categories. Sentiment analysis techniques can be used to assign a piece of text a single value that represents opinion expressed in that text. Sentiment analysis using lexicon approaches is already studied. Methods: A new approach to sentiment analysis using deep neural networks techniques is proposed. Deep neural networks using Sequence to sequence model is studied in this paper. The main objective of this paper is to identify the sequence of relationships among the words in the reviews. Customer reviews are taken from Amazon and sentiment analysis is done using the word embedding method. Results: The results obtained by the proposed method are compared with the baseline algorithms such as Naïve, and logistic regression. Conclusion: Confusion Matrix along with receiver operating characteristics and area under the curve is analyzed. The accuracy of the proposed methodology is compared with other algorithms.

Learning compact and discriminative hybrid neural network for dental caries classification

Dental caries, alternatively called tooth decay is considered as the most general and rapidly increasing diseases. According to the statistics, nearly 90% of the adults confront the issue of dental caries. Hence for proper dental care and health, early detection and diagnosis concerning the caries lesion plays a mandatory role. To achieve this, technique of Imaging Processing is being adopted which further helps the specialists to carry out precise diagnosis. Dental Caries can be categorized into different types depending on its position as well as seriousness. This categorization holds significant to carry out diagnosis and plan for the treatment of dental disease. The paper recommends the new HNN (Hybrid Neural Network) technique for detecting dental caries depending on, its type and therefore precisely classify the caries affected layer. HNN is a blend of ANN (Artificial Neural Network and DNN (Deep neural network) techniques. To obtain image classification, deep neural network relies upon the stacked sparse auto-encoder. Thereafter authentic information can be fetched from data via supervised fine-tuning and unsupervised pre-training. On the other hand ANN (Artificial neural network) relies upon the logistic regression for classifying the dental caries affected layer. Various processes involved in dental input image are as following: l. Pre-Processing, 2. Segmentation, 3. Feature Extraction and 4.Classification. Caries are categorized into 4 different layers by the HNN, namely: Enamel, Dentin, Pulp and root lesions. The proposed system yield in effective output against the input image provided and precisely categorizes the caries level. Hybrid neural network is the innovative technique which is giving good result in terms of accuracy and processing time.

An Integrated Imbalanced Learning and Deep Neural Network Model for Insider Threat Detection

The insider threat is a vital security problem concern in both the private and public sectors. A lot of approaches available for detecting and mitigating insider threats. However, the implementation of an effective system for insider threats detection is still a challenging task. In previous work, the Machine Learning (ML) technique was proposed in the insider threats detection domain since it has a promising solution for a better detection mechanism. Nonetheless, the (ML) techniques could be biased and less accurate when the dataset used is hugely imbalanced. Therefore, in this article, an integrated insider threat detection is named (AD-DNN), which is an integration of adaptive synthetic technique (ADASYN) sampling approach and deep neural network technique (DNN). In the proposed model (AD-DNN), the adaptive synthetic (ADASYN) is used to solve the imbalanced data issue and the deep neural network (DNN) for insider threat detection. The proposed model uses the CERT dataset for the evaluation process. The experimental results show that the proposed integrated model improves the overall detection performance of insider threats. A significant impact on the accuracy performance brings a better solution in the proposed model compared with the current insider threats detection system.