Automatic Classification Algorithm Research Articles

An Automatic Sleep Stage Classification Algorithm Using Improved Model Based Essence Features.

The automatic sleep stage classification technique can facilitate the diagnosis of sleep disorders and release the medical expert from labor-consumption work. In this paper, novel improved model based essence features (IMBEFs) were proposed combining locality energy (LE) and dual state space models (DSSMs) for automatic sleep stage detection on single-channel electroencephalograph (EEG) signals. Firstly, each EEG epoch is decomposed into low-level sub-bands (LSBs) and high-level sub-bands (HSBs) by wavelet packet decomposition (WPD), separately. Then, the DSSMs are estimated by the LSBs and the LE calculation is carried out on HSBs. Thirdly, the IMBEFs extracted from the DSSM and LE are fed into the appropriate classifier for sleep stage classification. The performance of the proposed method was evaluated on three public sleep databases. The experimental results show that under the Rechtschaffen’s and Kale’s (R&K) standard, the sleep stage classification accuracies of six classes on the Sleep EDF database and the Dreams Subjects database are 92.04% and 78.92%, respectively. Under the American Academy of Sleep Medicine (AASM) standard, the classification accuracies of five classes in the Dreams Subjects database and the ISRUC database reached 79.90% and 81.65%. The proposed method can be used for reliable sleep stage classification with high accuracy compared with state-of-the-art methods.

Learning dynamic simultaneous clustering and classification via automatic differential evolution and firework algorithm

Semi-supervised learning is significant data analysis method in the age of Big Data. The Bayesian-based classifier is a classical classification method in semi-supervised learning. Wherein, the classifier with clustering and classification technology have experienced the transformation from sequential structure to simultaneous structure. There are two main difficulties in simultaneous structure: the limited accuracy and diversity caused by rigid optimization algorithm; and the imbalance status between clustering and classification processes caused by insufficiently structure. To overcome these difficulties, a novel multi-objective differential evolution and firework algorithm for automatic simultaneous clustering and classification algorithm (MASCC-DE/FWA) is proposed. The main contributions of MASCC-DE/FWA contain: (1) Combination searching strategy for dynamic searching (2) Rapid and low-complexity silhouette coefficient as redesigned clustering objective function (3) Automatic clustering, opposition-based learning and adjusted mutual information for strengthening SCC-MOEA framework. The experimental result demonstrates that MASCC-DE/FWA performs better than other 8 state-of-art classification algorithms on synthetic dataset, 19 UCI datasets and image segmentation tasks.

Automatic text classification using machine learning and optimization algorithms

In the recent years, the volume of text documents in the form of digital way has grown up extremely in size. As significance, there is a need to be competent to automatically bring together and classify the documents based on their content. The main goal of text classification is to partition the unstructured set of documents into their respective categories based on its content. The main aim of this research work is to automatically classify the documents which are stored in the personal computer into their relevant categories. This work has two significant phases. In the first phase, the important features are selected for classification and the second phase is the classification of text documents. For selecting the optimal features, this research work proposes a new algorithm, optimization technique for feature selection (OTFS) algorithm. To estimate the proficiency of proposed feature selection algorithm, the OTFS algorithm was compared with the existing approaches artificial bee colony, firefly algorithm, ant colony optimization and particle swarm optimization. In the second phase, this research work proposed machine learning-based automatic text classification (MLearn-ATC) algorithm for text classification. In classification, the MLearn-ATC algorithm was compared with widely used classification techniques probabilistic neural network, support vector machine, K-nearest neighbor and Naive Bayes. From this, the output of first phase is used as the input for classification phase. The decisive results establish that the proposed algorithms achieve the better accuracy for optimizing the features and classifying the text documents based on their content.

Texture classification of fabric defects using machine learning

In this paper, a novel algorithm for automatic fabric defect classification was proposed, based on the combination of a texture analysis method and a support vector machine SVM. Three texture methods were used and compared, GLCM, LBP, and LPQ. They were combined with SVM’s classifier. The system has been tested using TILDA database. A comparative study of the performance and the running time of the three methods was carried out. The obtained results are interesting and show that LBP is the best method for recognition and classification and it proves that the SVM is a suitable classifier for such problems. We demonstrate that some defects are easier to classify than others.

Automatic Sleep Stage Detection: A Study on the Influence of Various PSG Input Signals.

Automatic sleep stage detection can be performed using a variety of input signals from a polysomnographic (PSG) recording. In this study, we investigate the effect of different input signals on the performance of feature-based automatic sleep stage classification algorithms with both a Random Forest (RF) and Multilayer Perceptron (MLP) classifier. Combinations of the EEG (electroencephalographic) signal and ECG (electrocardiographic), EMG (electromyographic) and respiratory signals as input are investigated as input with respect to using single channel and multi-channel EEG as input. The Physionet "You Snooze, You Win" dataset is used for the study. The RF classifier consistently outperforms our MLP implementation in all cases and is positively affected by specific signal combinations. The overall classification performance using a single channel EEG is high (an accuracy, precision and recall of 86.91 %, 89.52%, 86.91% respectively) using RF. The results are comparable to the performance obtained using six EEG channels as input. Adding respiratory signals to the inputs processed by RF increases the N2 stage detection performance with 20%, while adding the EMG signal improves the accuracy of the REM stage detection with 5%. Our analysis shows that adding specific signals as input to RF improves the accuracy of specific sleep stages and increases the overall performance. Using a combination of EEG and respiratory signals we achieved an accuracy of 93% for the RF classifier.

Electroencephalogram spike detection and classification by diagnosis with convolutional neural network

This work presents convolutional neural network (CNN) based methodology for electroencephalogram (EEG) classification by diagnosis: benign childhood epilepsy with centrotemporal spikes (rolandic epilepsy) (Group I) and structural focal epilepsy (Group II). Manual classification of these groups is sometimes difficult, especially, when no clinical record is available, thus presenting a need for an algorithm for automatic classification. The presented algorithm has the following steps: (i) EEG spike detection by morphological filter based algorithm; (ii) classification of EEG spikes using preprocessed EEG signal data from all channels in the vicinity of the spike detected; (iii) majority rule classifier application to all EEG spikes from a single patient. Classification based on majority rule allows us to achieve 80% average accuracy (despite the fact that from a single spike one would obtain only 58% accuracy).

Automated classification of cancer from fine needle aspiration cytological image use neural networks: A meta-analysis.

BackgroundThe role of retrospective analysis has been evolved greatly in cancer research. We undertook this meta‐analysis to evaluate the diagnostic value of Neural networks (NNs) in Fine needle aspiration cytological (FNAC) image of cancer.MethodsWe systematically retrieved 396 literatures on cytodiagnosis of NNs from Cochrane, PubMed, and EMBASE. After screening, only six studies were included in meta‐analysis finally. Data was comprehensively analyzed by RevMan and meta‐Disc software.ResultsA total of 1165 cases were extracted from six articles. Among them, 593 cases were in the abnormal/positive group and 572 cases in the normal/negative group. The pooled estimates for the NNs cytology were Area under ROC curve (AUC): 0.99, Sensitivity: 0.85 (95% CI:0.82‐0.88), Specificity: 0.96 (95% CI:0.94‐0.97), Positive Likelihood Ratio (LR):18.43 (95% CI:6.83‐49.74), Negative Likelihood Ratio (LR): 0.06 (95% CI:0.001‐0.58), and Diagnostic odds ratio (DOR): 343.21 (34.41‐3422.77).ConclusionsThis meta‐analysis confirms that NNs Automated Classification algorithm can facilitate to some extent the FNCA diagnosis of cancer.

The effects of visual congruence on increasing consumers’ brand engagement: An empirical investigation of influencer marketing on instagram using deep-learning algorithms for automatic image classification

Influencers are non-celebrity individuals who gain popularity on social media by posting visually attractive content (e.g., photos and videos) and by interacting with other users (i.e., Followers) to create a sense of authenticity and friendship. Brands partner with Influencers to garner engagement from their target consumers in a new marketing strategy known as “Influencer marketing.” Nonetheless, the theoretical underpinnings of such remains unknown. We suggest a new conceptual framework of “Visual-Congruence-induced Social Influence (VCSI),” which contextualizes the Similarity-Attraction Model in the Social Influence literature. Using VCSI, we delineate how Influencers use visual congruence as representations of shared interests in a specific area to build strong bonds with Followers. This intimate affiliation catalyzes (i.e., mediates) the positive effects of visual congruence on Followers’ brand engagement. To test these hypotheses, we conducted in vivo observations of Influencer marketing on Instagram. We collected >45,000 images and social media usage behaviors over 26 months. We then applied deep-learning algorithms to automatically classify each image and used social media analytics to disclose hidden associations between visual elements and brand engagement. Our hypothesis testing results provide empirical support for VCSI, advancing theories into the rapidly growing fields of multimodal content and Influencer marketing.

A-scan ultrasonic system for real time automatic cataract detection

Cataract is an ocular condition associated to the loss of the normal crystalline lens transparency, and its progression can result in a total loss of vision. The gold standard diagnostic method consists on qualitative observation through a slit lamp. This method has two limitations: incipient cataract may not be detected and the cataract hardness is subjectively evaluated. It may delay the diagnosis, or result on phacoemulsification surgeries complications when cataract hardness is not correctly estimated. On this study we present a new prototype for objective cataract detection and characterization, based on ultrasounds. The Eye Scan Ultrasonic System (ESUS) acquires A-scan signals at a nominal frequency of 20 MHz. The lens interfaces can be automatically detected in real time, based on the analysis of signal energy levels. The detection and characterization of cataract type and severity is done by an automatic classification algorithm based on features extracted in time and frequency domain. The system performance has been tested on preclinical data, and the beginning of clinical studies is expected shortly.

Automated ECG Waveform Annotation Based on Stacked Long Short-Term Memory

The classification of electrocardiogram (ECG) waveform segmentation techniques can be difficult due to physiological variation of heart rate and different characteristics of the different ECG waves in terms of shape, frequency, amplitude, and duration. The P-wave, PR-segment, QRS-complex, ST-segment, and T-wave are extracted as the feature for classification algorithm to diagnose specified cardiac disorders. This requires the implementation of algorithms that identify specific points within the ECG wave. Some previous computational algorithms for automatic classification of ECG segmentation are proposed to overcome limitations of manual inspection of the ECG. This study presents new insight into the ECG semantic segmentation problem is surmounted by a deep learning approach for automatic ECG wave-form. Long short-term memory (LSTM) is proposed for this task. This experimental study has been performed for six different waveforms of ECG signal that represents cardiac disorders obtained from the Physionet: QT database. Overall, LSTM performance achieved accuracy, sensitivity, specificity, precision, F1-score, is 93.36%, 86.85%, 95.78%, 81.79%, and 83.09%, respectively.

Automated ECG Waveform Annotation Based on Stacked Long Short-Term Memory

The classification of electrocardiogram (ECG) waveform segmentation techniques can be difficult due to physiological variation of heart rate and different characteristics of the different ECG waves in terms of shape, frequency, amplitude, and duration. The P-wave, PR-segment, QRS-complex, ST-segment, and T-wave are extracted as the feature for classification algorithm to diagnose specified cardiac disorders. This requires the implementation of algorithms that identify specific points within the ECG wave. Some previous computational algorithms for automatic classification of ECG segmentation are proposed to overcome limitations of manual inspection of the ECG. This study presents new insight into the ECG semantic segmentation problem is surmounted by a deep learning approach for automatic ECG wave-form. Long short-term memory (LSTM) is proposed for this task. This experimental study has been performed for six different waveforms of ECG signal that represents cardiac disorders obtained from the Physionet: QT database. Overall, LSTM performance achieved accuracy, sensitivity, specificity, precision, F1-score, is 93.36%, 86.85%, 95.78%, 81.79%, and 83.09%, respectively.

Artificial intelligence assisted bone lesion detection and classification in computed tomography scans of prostate cancer patients.

e17567 Background: Patients diagnosed with prostate cancer undergo computed tomography (CT) for pretreatment staging to rule out bone metastases. However, detection and classification of bone lesions on CT is challenging and subject to inter-reader variability. We present a cascaded deep learning algorithm for automatic detection and classification of bone lesions on staging CT in patients diagnosed with prostate cancer. Methods: CT scans from 56 patients with histopathologically proven prostate cancer were included. An expert radiologist annotated the extent of individual bone lesions (N = 4217) and labelled all regions as either benign or malignant. All scans were anonymized and normalized at the patient-level prior to training. Our method can be described as a two-stage framework, 1) A detection algorithm: Inspired by Yolo-v3 detection method, we designed a network with a backbone of darknet-53 pretrained on Coco dataset and four final scaling blocks to compensate for wide range of lesion diameters, 2) A classification algorithm: we formed a binary classifier based on ResNet-50 pretrained using the ImageNet dataset. We used a train/validation split equal to 90%/10% for this study. To facilitate the learning process, horizontal flipping, relative zooming and mean weighted averaging were used for data augmentation in stage 1. Instead, the classification algorithm took advantage of synthesized patches generated by Deep Convolutional Generative Adversarial Network (DC-GAN) for augmentation. Results: We could achieve a real-time (~120ms per slice) performance on our validation set with a median penalty of 0.3(0.02-0.78) false positives per true positive within each patient. Overall performance of our detection algorithm was 81% sensitivity and 86% positive predictive value. In stage 2, we obtained an accuracy of 89% for correct classification of benign from malignant bone lesions with no augmentation which was improved to 91% when we incorporated the augmented data for training. Conclusions: Our 2-stage algorithm sequentially detects and classifies bone lesions on CT of prostate cancer patients with a significant performance. To further improve our results and for generalizability we are accruing more data from different centers. Eventually, with greater dataset, both algorithms will be cascaded and trained as a whole unit to become one single tool for fully automatic detection and classification which serves as an aid for radiologists who read the staging CTs.

Automatic Detection and Classification of Road, Car, and Pedestrian Using Binocular Cameras in Traffic Scenes with a Common Framework

In order to solve the problems of traffic object detection, fuzzification, and simplification in real traffic environment, an automatic detection and classification algorithm for roads, vehicles, and pedestrians with multiple traffic objects under the same framework is proposed. We construct the final V view through a considerate U-V view method, which determines the location of the horizon and the initial contour of the road. Road detection results are obtained through error label reclassification, omitting point reassignment, and so an. We propose a peripheral envelope algorithm to determine sources of vehicles and pedestrians on the road. The initial segmentation results are determined by the regional growth of the source point through the minimum neighbor similarity algorithm. Vehicle detection results on the road are confirmed by combining disparity and color energy minimum algorithms with the object window aspect ratio threshold method. A method of multifeature fusion is presented to obtain the pedestrian target area, and the pedestrian detection results on the road are accurately segmented by combining the disparity neighbor similarity and the minimum energy algorithm. The algorithm is tested in three datasets of Enpeda, KITTI, and Daimler; then, the corresponding results prove the efficiency and accuracy of the proposed approach. Meanwhile, the real-time analysis of the algorithm is performed, and the average time efficiency is 13 pfs, which can realize the real-time performance of the detection process.

MAD saccade: statistically robust saccade threshold estimation via the median absolute deviation.

Saccade detection is a critical step in the analysis of gaze data. A common method for saccade detection is to use a simple threshold for velocity or acceleration values, which can be estimated from the data using the mean and standard deviation. However, this method has the downside of being influenced by the very signal it is trying to detect, the outlying velocities or accelerations that occur during saccades. We propose instead to use the median absolute deviation (MAD), a robust estimator of dispersion that is not influenced by outliers. We modify an algorithm proposed by Nyström and colleagues, and quantify saccade detection performance in both simulated and human data. Our modified algorithm shows a significant and marked improvement in saccade detection - showing both more true positives and less false negatives – especially under higher noise levels. We conclude that robust estimators can be widely adopted in other common, automatic gaze classification algorithms due to their ease of implementation.

Development of an algorithm for automatic classification of right ventricle deformation patterns in arrhythmogenic right ventricular cardiomyopathy.

BackgroundDifferent disease stages of arrhythmogenic right ventricular cardiomyopathy (ARVC) can be identified by right ventricle (RV) longitudinal deformation (strain) patterns. This requires assessment of the onset of shortening, (systolic) peak strain, and postsystolic index, which is time‐consuming and prone to inter‐ and intra‐observer variability. The aim of this study was to design and validate an algorithm to automatically classify RV deformation patterns.MethodsWe developed an algorithm based on specific local characteristics from the strain curves to detect the parameters required for classification. Determination of the onset of shortening by the algorithm was compared to manual determination by an experienced operator in a dataset containing 186 RV strain curves from 26 subjects carrying a pathogenic plakophilin‐2 (PKP2) mutation and 36 healthy subjects. Classification agreement between operator and algorithm was solely based on differences in onset shortening, as the remaining parameters required for classification of RV deformation patterns could be directly obtained from the strain curves.ResultsThe median difference between the onset of shortening determined by the experienced operator and by the automatic detector was 5.3 ms [inter‐quartile range (IQR) 2.7–8.6 ms]. 96% of the differences were within 1 time frame. Both methods correlated significantly with ρ = 0.97 (P < .001). For 26 PKP2 mutation carriers, there was 100% agreement in classification between the algorithm and experienced operator.ConclusionThe determination of the onset of shortening by the experienced operator was comparable to the algorithm. Our computer algorithm seems a promising method for the automatic classification of RV deformation patterns. The algorithm is publicly available at the MathWorks File Exchange.

Model-based unsupervised clustering for distinguishing Cuvier's and Gervais' beaked whales in acoustic data

Passive acoustic monitoring (PAM), particularly autonomous platforms, offers many advantages in monitoring phonating deep-diving marine mammals in oceanic environment. Relevant data can be obtained day and night continuously over long durations and in any weather conditions. It provides a cost-efficient solution with greater detection ranges when compared to traditional large research vessel and aerial visual surveys requiring keeping expert observers on station for long periods of time and relying on good visibility and calm seas. Therefore, PAM is becoming a preferred tool to assess population dynamics trends and health of deep-water marine mammal stocks. However the large volumes of collected data require robust automatic detection and classification algorithms to identify marine mammals in recordings. As for beaked whales, one of the challenging automatic processing goals is the identification of different species to advance our understanding of their role in the marine ecosystem. At present, traditional detection and classification methods employ searches for acoustic events above a user-defined signal-to-noise ratio threshold in the frequency band of interest and further rely on an experienced operator's manual inspection for species classification and removal of false positives. Current passive monitoring data collection systems yield large volumes of acoustic data, therefore a manual classification approach becomes very time-consuming and impractical. This paper focuses on developing a multi-stage automatic classifier for beaked whale species. The proposed method utilizes unsupervised machine clustering of signal attributes extracted from potential detection events flagged by an energy-band detector. The proposed algorithm was benchmarked against a manually annotated workshop dataset and applied to acoustic data collected in the northern Gulf of Mexico. The algorithm classifies beaked whale species in automatic mode with minimal operator involvement only at the validation stage. When compared with the manually annotated classification dataset, the proposed method achieved a recall rate of 82.8% for Cuvier's and 77.9% for Gervais' species in automatic mode. New insights on habitat use by different species of beaked whales in the Gulf of Mexico were gained when using the species-specific classifier. The high spatial resolution acoustic monitoring results showed that the habitat preferences of two dominating beaked whale species in the Gulf of Mexico support the habitat division (ecological niche) hypothesis.

Ballistic Labeling of Pyramidal Neurons in Brain Slices and in Primary Cell Culture.

It has been reported that the size and shape of dendritic spines is related to their structural plasticity. To identify the morphological structure of pyramidal neurons and dendritic spines, a ballistic labeling technique can be utilized. In the present protocol, pyramidal neurons are labeled with DilC18(3) dye and analyzed using neuronal reconstruction software to assess neuronal morphology and dendritic spines. To investigate neuronal structure, dendritic branching analysis and Sholl analysis are performed, allowing researchers to draw inferences about dendritic branching complexity and neuronal arbor complexity, respectively. The evaluation of dendritic spines is conducted using an automatic assisted classification algorithm integral to the reconstruction software, which classifies spines into four categories (i.e., thin, mushroom, stubby, filopodia). Furthermore, an additional three parameters (i.e., length, head diameter, and volume) are also chosen to assess alterations in dendritic spine morphology. To validate the potential of wide application of the ballistic labeling technique, pyramidal neurons from in vitro cell culture were successfully labeled. Overall, the ballistic labeling method is unique and useful for visualizing neurons in different brain regions in rats, which in combination with sophisticated reconstruction software, allows researchers to elucidate the possible mechanisms underlying neurocognitive dysfunction.

Low overhead NOMA receiver with automatic modulation classification techniques

Recently non-orthogonal multiple access (NOMA) technique has been proposed for boosting spectral efficiencies in 5G cellular networks. In NOMA the transmitter maps a predefined number of user symbols on to the same resources. The conventional receiver for NOMA is the well-known serial interference cancellation. To overcome the problem of high overhead signalling, a receiver with an automatic modulation classification (AMC) algorithm for a K-user NOMA which identifies the modulation scheme of users before symbol detection is proposed. First, a two-user NOMA (TU-NOMA) scheme is treated and the receiver is extended to more general cases. For evaluating the performance of the proposed receiver its bit error rate (BER) for TU-NOMA case has been driven and shown that this can be approximated as the sum of the demodulator BER and the AMC algorithm error probability. The simulation results show the BER degradation can be neglected if an enough number of symbols with sufficiently large SNR values are available.

Brain MRI Classification using Deep Learning Algorithm

The brain tumor is one of the most dangerous, common and aggressive diseases which leads to a very short life expectancy at the highest grade. Thus, to prevent life from such disease, early recognition, and fast treatment is an essential step. In this approach, MRI images are used to analyze brain abnormalities. The manual investigation of brain tumor classification is a time-consuming task and there might have possibilities of human errors. Hence accurate analysis in a tiny span of time is an essential requirement. In this approach, the automatic brain tumor classification algorithm using a highly accurate Convolutional Neural Network (CNN) algorithm is presented. Initially, the brain part is segmented by thresholding approach followed by a morphological operation. The AlexNet transfer learning network of CNN is used because of the limitation of the brain MRI dataset. The classification layer of Alexnet is replaced by the softmax layer with benign and malignant training images and trained using small weights. The experimental analysis demonstrates that the proposed system achieves the F-measure of 98.44% with low complexity than the state-of-arts method.

A comparison of clustering algorithms for automatic modulation classification

In this paper, the k-means, k-medoids, fuzzy c-means, Density-Based Spatial Clustering of Applications with Noise (DBSCAN), Ordering Points To Identify the Clustering Structure (OPTICS), and hierarchical clustering algorithms (with the addition of the elbow method) are examined for the purpose of Automatic Modulation Classification (AMC). This study compares these algorithms in terms of classification accuracy and execution time for either estimating the modulation order, determining centroid locations, or both. The best performing algorithms are combined to provide a simple AMC method which is then evaluated in an Additive White Gaussian Noise (AWGN) channel with M-Quadrature Amplitude Modulation (QAM) and M-Phase Shift Keying (PSK). Such an AMC method does not rely on any thresholds to be set by a human or machine learning algorithm, resulting in a highly flexible system. The proposed method can be configured to not give false positives, making it suitable for applications such as spectrum monitoring and regulatory enforcement.