Deep Learning-based Classification Model Research Articles

Machine Learning and Deep Learning Based Computational Techniques in Automatic Agricultural Diseases Detection: Methodologies, Applications, and Challenges

Plant disease detection is a critical issue that needs to be focused on for productive agriculture and economy. Detecting plant disease using traditional methods is a tedious job as it requires a tremendous amount of work, time, and expertise. Automatic plant disease detection is an important research area that has recently gained a lot of attention among the academicians, researchers, and practitioners. Machine Learning and Deep Learning can help identify the plant disease at the initial stage as soon as it appears on plant leaves. In this state-of-an-art review, a thorough investigation has been performed to evaluate the possibility of using Machine Learning models to identify plant diseases. In this study, diseases and infections of four types of crops, i.e., Tomato, Rice, Potato, and Apple, are considered. Initially, numerous possible infections and diseases on these four kinds of crops are studied along with their reason for the occurrence and possible symptoms for their detections. An in-depth study of the different steps involved in plant disease detection and classification using Machine Learning and Deep Learning is provided. Various datasets available online for plant disease detection have also been presented. Along with this, a detailed study on various existing Machine Learning and Deep Learning-based classification models proposed by different researchers across the world for four considered crops in terms of their performance evaluations, the dataset used, and the feature extraction method is discussed. At last, various challenges in the use of machine learning and deep learning for plant disease detection and future research directions are enumerated and presented.

Epilepsy attacks recognition based on 1D octal pattern, wavelet transform and EEG signals

Electroencephalogram (EEG) signals have been generally utilized for diagnostic systems. Nowadays artificial intelligence-based systems have been proposed to classify EEG signals to ease diagnosis process. However, machine learning models have generally been used deep learning based classification model to reach high classification accuracies. This work focuses classification epilepsy attacks using EEG signals with a lightweight and simple classification model. Hence, an automated EEG classification model is presented. The used phases of the presented automated EEG classification model are (i) multileveled feature generation using one-dimensional (1D) octal-pattern (OP) and discrete wavelet transform (DWT). Here, main feature generation function is the presented octal-pattern. DWT is employed for level creation. By employing DWT frequency coefficients of the EEG signal is obtained and octal-pattern generates texture features from raw EEG signal and wavelet coefficients. This DWT and octal-pattern based feature generator extracts 128 × 8 = 1024 (Octal-pattern generates 128 features from a signal, 8 signal are used in the feature generation 1 raw EEG and 7 wavelet low-pass filter coefficients). (ii) To select the most useful features, neighborhood component analysis (NCA) is deployed and 128 features are selected. (iii) The selected features are feed to k nearest neighborhood classifier. To test this model, an epilepsy seizure dataset is used and 96.0% accuracy is attained for five categories. The results clearly denoted the success of the presented octal-pattern based epilepsy classification model.

DCBRTS: A Classification-Summarization Approach for Evolving Tweet Streams in Multiobjective Optimization Framework

The emergence of social media platforms like Twitter has become a prominent communication source in disaster outbreak. NGOs, Government agencies leverage twitter’s open and public features to provide immediate relief. Nevertheless, situational information gets immersed in millions of tweets with varying characteristics. Examining each tweet can be cumbersome and time-consuming. Thus, the efficient extraction of disaster-related situational tweets and getting information from all the extracted tweets is required. In the current paper, we have developed a novel framework that uses a deep learning-based classification model to separate the situational tweets from others and summarize them in real-time. Our system is a three-phase process: (a) Creating tweet clusters using a representative set of tweets from the initial set of extracted tweets using a multi-objective optimization concept; (b) When a new tweet arrives, the clusters are updated. The new tweet is classified as situational vs. non-situational. If situational, it is assigned to the closest cluster or new cluster. This assignment is based on its weighted average of syntactic and semantic distances and relevancy to the cluster; (c) Summary is formulated by extracting tweets from each cluster. The proposed approach’s superior performance on four datasets related to different disaster-related events indicates the developed framework’s efficiency over the state-of-the-art techniques.

Radar-Spectrogram-Based UAV Classification Using Convolutional Neural Networks.

With the upsurge in the use of Unmanned Aerial Vehicles (UAVs) in various fields, detecting and identifying them in real-time are becoming important topics. However, the identification of UAVs is difficult due to their characteristics such as low altitude, slow speed, and small radar cross-section (LSS). With the existing deterministic approach, the algorithm becomes complex and requires a large number of computations, making it unsuitable for real-time systems. Hence, effective alternatives enabling real-time identification of these new threats are needed. Deep learning-based classification models learn features from data by themselves and have shown outstanding performance in computer vision tasks. In this paper, we propose a deep learning-based classification model that learns the micro-Doppler signatures (MDS) of targets represented on radar spectrogram images. To enable this, first, we recorded five LSS targets (three types of UAVs and two different types of human activities) with a frequency modulated continuous wave (FMCW) radar in various scenarios. Then, we converted signals into spectrograms in the form of images by Short time Fourier transform (STFT). After the data refinement and augmentation, we made our own radar spectrogram dataset. Secondly, we analyzed characteristics of the radar spectrogram dataset with the ResNet-18 model and designed the ResNet-SP model with less computation, higher accuracy and stability based on the ResNet-18 model. The results show that the proposed ResNet-SP has a training time of 242 s and an accuracy of 83.39%, which is superior to the ResNet-18 that takes 640 s for training with an accuracy of 79.88%.

EEG-Based Estimation on the Reduction of Negative Emotions for Illustrated Surgical Images.

Electroencephalogram (EEG) biosignals are widely used to measure human emotional reactions. The recent progress of deep learning-based classification models has improved the accuracy of emotion recognition in EEG signals. We apply a deep learning-based emotion recognition model from EEG biosignals to prove that illustrated surgical images reduce the negative emotional reactions that the photographic surgical images generate. The strong negative emotional reactions caused by surgical images, which show the internal structure of the human body (including blood, flesh, muscle, fatty tissue, and bone) act as an obstacle in explaining the images to patients or communicating with the images with non-professional people. We claim that the negative emotional reactions generated by illustrated surgical images are less severe than those caused by raw surgical images. To demonstrate the difference in emotional reaction, we produce several illustrated surgical images from photographs and measure the emotional reactions they engender using EEG biosignals; a deep learning-based emotion recognition model is applied to extract emotional reactions. Through this experiment, we show that the negative emotional reactions associated with photographic surgical images are much higher than those caused by illustrated versions of identical images. We further execute a self-assessed user survey to prove that the emotions recognized from EEG signals effectively represent user-annotated emotions.

An Ensemble Deep Neural Network Model for Onion-Routed Traffic Detection to Boost Cloud Security

Anonymous network communication using onion routing networks such as Tor are used to guard the privacy of sender by encrypting all messages in the overlapped network. These days most of the onion routed communications are not only used for decent cause but also cyber offenders are ill-using onion routings for scanning the ports, hacking, exfiltration of theft data, and other types of online frauds. These cyber-crime attempts are very vulnerable for cloud security. Deep learning is highly effective machine learning method for prediction and classification. Ensembling multiple models is an influential approach to increase the efficiency of learning models. In this work, an ensemble deep learning-based classification model is proposed to detect communication through Tor and non-Tor network. Three different deep learning models are combined to achieve the ensemble model. The proposed model is also compared with other machine learning models. Classification results shows the superiority of the proposed model than other models.

DOMAIN ADAPTATION WITH CYCLEGAN FOR CHANGE DETECTION IN THE AMAZON FOREST

Abstract. Deep learning classification models require large amounts of labeled training data to perform properly, but the production of reference data for most Earth observation applications is a labor intensive, costly process. In that sense, transfer learning is an option to mitigate the demand for labeled data. In many remote sensing applications, however, the accuracy of a deep learning-based classification model trained with a specific dataset drops significantly when it is tested on a different dataset, even after fine-tuning. In general, this behavior can be credited to the domain shift phenomenon. In remote sensing applications, domain shift can be associated with changes in the environmental conditions during the acquisition of new data, variations of objects’ appearances, geographical variability and different sensor properties, among other aspects. In recent years, deep learning-based domain adaptation techniques have been used to alleviate the domain shift problem. Recent improvements in domain adaptation technology rely on techniques based on Generative Adversarial Networks (GANs), such as the Cycle-Consistent Generative Adversarial Network (CycleGAN), which adapts images across different domains by learning nonlinear mapping functions between the domains. In this work, we exploit the CycleGAN approach for domain adaptation in a particular change detection application, namely, deforestation detection in the Amazon forest. Experimental results indicate that the proposed approach is capable of alleviating the effects associated with domain shift in the context of the target application.

Predicting the invasiveness of lung adenocarcinomas appearing as ground-glass nodule on CT scan using multi-task learning and deep radiomics.

BackgroundDue to different treatment method and prognosis of different subtypes of lung adenocarcinomas appearing as ground-glass nodules (GGNs) on computed tomography (CT) scan, it is important to classify invasive adenocarcinomas from non-invasive adenocarcinomas. The purpose of this paper is to build and evaluate the performance of deep learning networks on the differentiation the invasiveness of lung adenocarcinoma appearing as GGNs.MethodsThis retrospective study included 886 GGNs from 794 pathological confirmed patients with lung adenocarcinoma for training and testing the proposed networks. Three deep learning networks, namely XimaNet (deep learning-based classification model), XimaSharp (classification and nodule segmentation model), and Deep-RadNet (deep learning and radiomics combined classification model, i.e., deep radiomics) were built. Three classification tasks, namely task 1: classification of AAH/AIS and MIA, task 2: classification of MIA and IAC, and task 3: classification of non-invasive adenocarcinomas and invasive adenocarcinomas (AAH/AIS&MIA and IAC) were conducted to evaluate the model performance. The Z-test was used to compare the model performance.ResultsThe AUC for classification of AAH/AIS with MIA were 0.891, 0.841 and 0.779 for Deep-RadNet, XimaNet and XimaSharp respectively. The AUC for classification of MIA with IAC were 0.889, 0.785 and 0.778 for three networks and AUC for classification of AAH/AIS&MIA with IAC were 0.941, 0.892 and 0.827 respectively. The performance of deep_RadNet was better than the other two models with the Z-test (P<0.05).ConclusionsDeep-RadNet with the visual heat map could evaluate the invasiveness of GGNs accurately and intuitively, providing a theoretical basis for individualized and accurate medical treatment of patients with GGNs.

Deep learning-based anatomical site classification for upper gastrointestinal endoscopy

PurposeUpper gastrointestinal (GI) endoscopic image documentation has provided an efficient, low-cost solution to address quality control for endoscopic reporting. The problem is, however, challenging for computer-assisted techniques, because different sites have similar appearances. Additionally, across different patients, site appearance variation may be large and inconsistent. Therefore, according to the British and modified Japanese guidelines, we propose a set of oesophagogastroduodenoscopy (EGD) images to be routinely captured and evaluate its efficiency for deep learning-based classification methods.MethodsA novel EGD image dataset standardising upper GI endoscopy to several steps is established following landmarks proposed in guidelines and annotated by an expert clinician. To demonstrate the discrimination of proposed landmarks that enable the generation of an automated endoscopic report, we train several deep learning-based classification models utilising the well-annotated images.ResultsWe report results for a clinical dataset composed of 211 patients (comprising a total of 3704 EGD images) acquired during routine upper GI endoscopic examinations. We find close agreement between predicted labels using our method and the ground truth labelled by human experts. We observe the limitation of current static image classification scheme for EGD image classification.ConclusionOur study presents a framework for developing automated EGD reports using deep learning. We demonstrate that our method is feasible to address EGD image classification and can lead towards improved performance and additionally qualitatively demonstrate its performance on our dataset.

Optimal Deep Learning based Classification Model for Mitral Valve Diagnosis System

In present days, the domain of mitral valve (MV) diagnosis so common due to the changing lifestyle in day to day life. The increased number of MV disease necessitates the development of automated disease diagnosis model based on segmentation and classification. This paper makes use of deep learning (DL) model to develop a MV classification model to diagnose the severity level. For the accurate classification of ML, this paper applies the DL model called convolution neural network (CNN-MV) model. And, an edge detection based segmentation model is also applied which will helps to further enhance the performance of the classifier. Due to the non-availability of MV dataset, we have collected a MV dataset of our own from a total of 211 instances. A set of three validation parameters namely accuracy, sensitivity and specificity are applied to indicate the effective operation of the CNN-MV model. The obtained simulation outcome pointed out that the presented CNN-MV model functions as an appropriate tool for MV diagnosis.

Deep learning-based classification model for botnet attack detection

Botnets are vectors through which hackers can seize control of multiple systems and conduct malicious activities. Researchers have proposed multiple solutions to detect and identify botnets in real time. However, these proposed solutions have difficulties in keeping pace with the rapid evolution of botnets. This paper proposes a model for detecting botnets using deep learning to identify zero-day botnet attacks in real time. The proposed model is trained and evaluated on a CTU-13 dataset with multiple neural network designs and hidden layers. Results demonstrate that the deep-learning artificial neural network model can accurately and efficiently identify botnets.

Active Balancing Mechanism for Imbalanced Medical Data in Deep Learning–Based Classification Models

Imbalanced data always has a serious impact on a predictive model, and most under-sampling techniques consume more time and suffer from loss of samples containing critical information during imbalanced data processing, especially in the biomedical field. To solve these problems, we developed an active balancing mechanism (ABM) based on valuable information contained in the biomedical data. ABM adopts the Gaussian naïve Bayes method to estimate the object samples and entropy as a query function to evaluate sample information and only retains valuable samples of the majority class to achieve under-sampling. The Physikalisch Technische Bundesanstalt diagnostic electrocardiogram (ECG) database, including 5,173 normal ECG samples and 26,654 myocardial infarction ECG samples, is applied to verify the validity of ABM. At imbalance rates of 13 and 5, experimental results reveal that ABM takes 7.7 seconds and 13.2 seconds, respectively. Both results are significantly faster than five conventional under-sampling methods. In addition, at the imbalance rate of 13, ABM-based data obtained the highest accuracy of 92.23% and 97.52% using support vector machines and modified convolutional neural networks (MCNNs) with eight layers, respectively. At the imbalance rate of 5, the processed data by ABM also achieved the best accuracy of 92.31% and 98.46% based on support vector machines and MCNNs, respectively. Furthermore, ABM has better performance than two compared methods in F 1-measure, G-means, and area under the curve. Consequently, ABM could be a useful and effective approach to deal with imbalanced data in general, particularly biomedical myocardial infarction ECG datasets, and the MCNN can also achieve higher performance compared to the state of the art.

A Novel Virtual Sample Generation Method to Overcome the Small Sample Size Problem in Computer Aided Medical Diagnosing

Deep neural networks are successful learning tools for building nonlinear models. However, a robust deep learning-based classification model needs a large dataset. Indeed, these models are often unstable when they use small datasets. To solve this issue, which is particularly critical in light of the possible clinical applications of these predictive models, researchers have developed approaches such as virtual sample generation. Virtual sample generation significantly improves learning and classification performance when working with small samples. The main objective of this study is to evaluate the ability of the proposed virtual sample generation to overcome the small sample size problem, which is a feature of the automated detection of a neurodevelopmental disorder, namely autism spectrum disorder. Results show that our method enhances diagnostic accuracy from 84%–95% using virtual samples generated on the basis of five actual clinical samples. The present findings show the feasibility of using the proposed technique to improve classification performance even in cases of clinical samples of limited size. Accounting for concerns in relation to small sample sizes, our technique represents a meaningful step forward in terms of pattern recognition methodology, particularly when it is applied to diagnostic classifications of neurodevelopmental disorders. Besides, the proposed technique has been tested with other available benchmark datasets. The experimental outcomes showed that the accuracy of the classification that used virtual samples was superior to the one that used original training data without virtual samples.

Deep learning model for unstructured knowledge classification using structural features

Automatic text classification is widely used as the basic method for analyzing data. While classification methods like the support vector machine (SVM) have exhibited impressive performance in the area, the recent use of deep learning has led to considerable progress in text classification. This study proposes a deep learning–based classification model called DEEP-I to classify information on national research and development with complex structural features, a large amount of text, and large-scale classes. In addition to the word–sentence structure of a simple document, the number of stacking layers of the deep model is increased in light of the higher-level structure of the items. Experiments on 180,000 datasets and 366 classification schemes showed that the proposed model can improve classification performance by 22.7% over the traditional SVM and 15.7% over a deep learning model that uses only structured features of word sentences. This improvement was achieved because the multi-layered stacking method was applied to enhance learning by increasing depth by five to 10 times that of the conventional deep learning model and effectively combining features of heterogeneous items. The proposed model is also applicable to datasets containing documents with complex structures.

Deep Learning for Smart Industry: Efficient Manufacture Inspection System With Fog Computing

With the rapid development of Internet of things devices and network infrastructure, there have been a lot of sensors adopted in the industrial productions, resulting in a large size of data. One of the most popular examples is the manufacture inspection, which is to detect the defects of the products. In order to implement a robust inspection system with higher accuracy, we propose a deep learning based classification model in this paper, which can find the possible defective products. As there may be many assembly lines in one factory, one huge problem in this scenario is how to process such big data in real time. Therefore, we design our system with the concept of fog computing. By offloading the computation burden from the central server to the fog nodes, the system obtains the ability to deal with extremely large data. There are two obvious advantages in our system. The first one is that we adapt the convolutional neural network model to the fog computing environment, which significantly improves its computing efficiency. The other one is that we work out an inspection model, which can simultaneously indicate the defect type and its degree. The experiments well prove that the proposed method is robust and efficient.