Obtained Classification Accuracy Research Articles

A Classification method for Insects using Data Augmentation and Deep Neural Networks

In the nature, there exists a huge number of species of insects. Insects have caused damage for human and crops. Traditional identification methods of insects require expert knowledge and time consuming. Therefore, the automatic identification and classification have been more and more necessary. In recent year, one of the efficient approaches to classify insects is the application of Deep neural networks. The paper presents the improvements for the classification of insect images. Firstly, we apply the data augmentation to improve the number of insect images. Then, we applied various deep neural networks to improve the classification accuracy of insect classification. Obtained classification accuracy of 98% on public insect datasets shows the efficiency of the proposed method.

An adversarial learning approach to generate pressure support ventilation waveforms for asynchrony detection

Background and objectiveMechanical ventilation is a life-saving treatment for critically-ill patients. During treatment, patient-ventilator asynchrony (PVA) can occur, which can lead to pulmonary damage, complications, and higher mortality. While traditional detection methods for PVAs rely on visual inspection by clinicians, in recent years, machine learning models are being developed to detect PVAs automatically. However, training these models requires large labeled datasets, which are difficult to obtain, as labeling is a labour-intensive and time-consuming task, requiring clinical expertise. Simulating the lung-ventilator interactions has been proposed to obtain large labeled datasets to train machine learning classifiers. However, the obtained data lacks the influence of different hardware, of servo-controlled algorithms, and different sources of noise. Here, we propose VentGAN, an adversarial learning approach to improve simulated data by learning the ventilator fingerprints from unlabeled clinical data. MethodsIn VentGAN, the loss functions are designed to add characteristics of clinical waveforms to the generated results, while preserving the labels of the simulated waveforms. To validate VentGAN, we compare the performance for detection and classification of PVAs when training a previously developed machine learning algorithm with the original simulated data and with the data generated by VentGAN. Testing is performed on independent clinical data labeled by experts. The McNemar test is applied to evaluate statistical differences in the obtained classification accuracy. ResultsVentGAN significantly improves the classification accuracy for late cycling, early cycling and normal breaths (p< 0.01); no significant difference in accuracy was observed for delayed inspirations (p = 0.2), while the accuracy decreased for ineffective efforts (p< 0.01). ConclusionsGeneration of realistic synthetic data with labels by the proposed framework is feasible and represents a promising avenue for improving training of machine learning models.

Hybrid optical convolutional neural network with convolution kernels trained in the spatial domain

Although many hybrid convolutional neural networks using optical convolution processors have been demonstrated for high energy efficiency and fast image processing speed, the need for accurate modeling of the optical convolution hardware and large Fourier kernel sizes result in a long and energy-intensive training process. Here, we demonstrate a hybrid convolutional neural network based on an optimized optical convolution processor—the system uses kernels trained in the spatial domain and compensates the optical path mismatch that arises from the reflection of digital micromirror devices for convolution computation. The spatial-domain convolution kernels are Fourier transformed and then binarized before being programmed into a digital micromirror device in the optical convolution processor. This method minimizes overhead from the optical convolution process in the training of the hybrid convolutional neural network, and it can improve the energy efficiency and reduce the training time of the hybrid optical convolutional neural network. For convolution computation using digital micromirror devices with grayscale images, bit-plane-wise convolution is also presented. We tested the hybrid neural network on MNIST, Fashion-MNIST, and CIFAR-10 color datasets and obtained classification accuracies of 98.7%, 85.8%, and 57.8%, respectively.

Efficient Deep Learning model for de-husked Areca nut classification

Areca nut is a widely used agricultural product in India and even over the globe. Areca nut, a fruit of areca palm (Areca catechu) is grown widely in the Asia-Pacific region.. Areca nut segregation is of prime importance in the areca nut industry. The quality segregation of peeled/de-husked nuts requires skilled workers. This process of manual segregation is time-consuming and can lead to erroneous classification. Recent deep learning (DL) advances have improved the performance in multi-class problems. The present work presents the classification of de-husked areca nut among five classes using an efficient deep learning customized Convolutional Neural Network (CNN) and the results of this model were compared with the standard AlexNet architecture. The new CNN model was customized to obtain classification accuracy higher than the existing ones. A dataset of 300 nuts (60 per class) was created using a specially designed instrumentation setup. The areca nut images were then pre-processed and fed to these models to learn the features of the areca nut from different classes. The confusion matrix and Area Under the Curve - Receiver Operating Characteristics (AUC- ROC) were employed to assess the results of these models and cross-validated with 5 and 10-fold. The experimental results show that the CNN outperformed the AlexNet model with an average accuracy of 97.33% and 98.34%, F1 score of 97.48%, and 98.45% for 5 and 10 folds, respectively.

3DMASC: Accessible, explainable 3D point clouds classification. Application to bi-spectral topo-bathymetric lidar data

Three-dimensional data have become increasingly present in earth observation over the last decades. However, many 3D surveys are still underexploited due to the lack of accessible and explainable automatic classification methods, for example, new topo-bathymetric lidar data. In this work, we introduce explainable machine learning for 3D data classification using Multiple Attributes, Scales, and Clouds under 3DMASC, a new workflow. This workflow introduces multi-cloud classification through dual-cloud features, encrypting local spectral and geometrical ratios and differences. 3DMASC uses classical multi-scale descriptors adapted to all types of 3D point clouds and new ones based on their spatial variations. In this paper, we present the performances of 3DMASC for multi-class classification of topo-bathymetric lidar data in coastal and fluvial environments. We show how multivariate and embedded feature selection allows the building of optimized predictor sets of reduced complexity, and we identify features particularly relevant for coastal and riverine scene descriptions. Our results show the importance of dual-cloud features, lidar return-based attributes averaged over specific scales, and of statistics of dimensionality-based and spectral features. Additionally, they indicate that small to medium spherical neighbourhood diameters (<7 m) are sufficient to build effective classifiers, namely when combined with distance-to-ground or distance-to-water-surface features. Without using optional RGB information, and with a maximum of 37 descriptors, we obtain classification accuracies between 91 % for complex multi-class tasks and 98 % for lower-level processing using models trained on less than 2000 samples per class. Comparisons with classical point cloud classification methods show that 3DMASC features have a significantly improved descriptive power. Our contributions are made available through a plugin in the CloudCompare software, allowing non-specialist users to create classifiers for any type of 3D data characterized by 1 or 2 point clouds (airborne or terrestrial lidar, structure from motion), and two labelled topo-bathymetric lidar datasets, available on https://opentopography.org/.

Chirplet transform based time frequency analysis of speech signal for automated speech emotion recognition

Nowadays, the recognition of emotion using the speech signal has gained popularity because of its vast number of applications in different fields like medicine, online marketing, online search engines, the education system, criminal investigations, traffic collisions, etc. Many researchers have adopted different methodologies to improve emotion classification accuracy using speech signals. In our study, time–frequency (TF) analysis-based features were used to analyze the emotion classification performance. We used a novel TF analysis method called the chirplet transform (CT) to find the TF matrix of the speech signal. We then calculated the proposed TF-based permutation entropy (TFPE) feature using the TF matrix of the speech signal. To reduce the feature dimension and select the most informative emotional feature, we employed the genetic algorithm (GA) feature selection method. Then, the selected TFPE features are used as input to machine learning classifiers such as SVM, RF, DT, and KNN to classify the emotions in the speech signal. We obtained classification accuracy of 77.2%, 69.57%, 68.78%, 56.9%, and 99.1% for the EMO-DB, EMOVO, SAVEE, IEMOCAP, and TESS datasets without the GA feature selection method. The emotion classification accuracy increased to 85.6%, 78.33%, 77.76%, 63.15%, and 100% with the GA feature selection method. We compared our results with other methods and found that our method performed better in emotion classification than the state-of-the-art methods.

The LSST AGN Data Challenge: Selection Methods

Development of the Rubin Observatory Legacy Survey of Space and Time (LSST) includes a series of Data Challenges (DCs) arranged by various LSST Scientific Collaborations that are taking place during the project's preoperational phase. The AGN Science Collaboration Data Challenge (AGNSC-DC) is a partial prototype of the expected LSST data on active galactic nuclei (AGNs), aimed at validating machine learning approaches for AGN selection and characterization in large surveys like LSST. The AGNSC-DC took place in 2021, focusing on accuracy, robustness, and scalability. The training and the blinded data sets were constructed to mimic the future LSST release catalogs using the data from the Sloan Digital Sky Survey Stripe 82 region and the XMM-Newton Large Scale Structure Survey region. Data features were divided into astrometry, photometry, color, morphology, redshift, and class label with the addition of variability features and images. We present the results of four submitted solutions to DCs using both classical and machine learning methods. We systematically test the performance of supervised models (support vector machine, random forest, extreme gradient boosting, artificial neural network, convolutional neural network) and unsupervised ones (deep embedding clustering) when applied to the problem of classifying/clustering sources as stars, galaxies, or AGNs. We obtained classification accuracy of 97.5% for supervised models and clustering accuracy of 96.0% for unsupervised ones and 95.0% with a classic approach for a blinded data set. We find that variability features significantly improve the accuracy of the trained models, and correlation analysis among different bands enables a fast and inexpensive first-order selection of quasar candidates.

Fast channel selection method using modified camel travelling behavior algorithm

Brain computer interface (BCI) is a protocol to communicate between the human brain and a device or application using brain signals. These signals translated to useful commands by using features extraction and classification. The most widely used features is the power of alpha and beta rhythms. This type of features gives only 70% of classification accuracy without any extra processing using fixed channels to read the signals. Because the distribution of the power in the brain is not a standard for all people, each one has his own brain power map. A selection algorithm is used to find the best channels that could generate higher power than the fixed ones. Modified camel travelling behavior algorithm is used to select the channels that used to extract the power of alpha and beta bands of motor imagery signals. This algorithm is faster to find the best set of channels, and obtain classification accuracy more than 95% using support vector machine classifier.

Detection of antioxidants in edible oil by two-dimensional correlation spectroscopy combined with convolutional neural network

In this study, the qualitative identification and quantitative prediction of antioxidants in edible oil were achieved by the two-dimensional correlation spectroscopy combined with convolutional neural network (CNN). The CNN model obtained classification accuracy of 97.56%, which is significantly higher than that of the partial least squares discriminant analysis (PLS-DA) model. Comparing the performance of the established partial least squares (PLS), CNN, and CNN combined with PLS (CNN_PLS) regression models, it is found that the CNN model has the best quantitative performance among the three regression models, with a root mean square errors for prediction (RMSEP) of 0.0232 and a coefficient of determination for prediction (R2p) of 0.9703. At the same time, it is found that the prediction ability of the CNN_PLS model is affected by the feature extraction ability of CNN models and the output dimension of convolution layers in CNN models. The results demonstrate the potential of the proposed method for the detection of antioxidant in edible oils.

Effective Feature Selection Strategy for Supervised Classification based on an Improved Binary Aquila Optimization Algorithm

Feature Selection (FS) is considered a crucial step in machine learning and data mining tasks, which facilitates minimizing the direct consequence of redundant and irrelevant attributes on the model’s accuracy. Hence, the researchers developed various algorithms to choose the most appropriate features for improving the accuracy rate of the presented dataset. Nevertheless, these algorithms may fall into local optima problems when applied to substantial feature sizes’ datasets. In this paper, for handling the FS strategy through dimensionality-lessening while improving the classification accuracy, an effective Aquila Optimization (AO) algorithm is introduced. AO has stable exploration and exploitation capabilities. It is enhanced by integrating a random position amendment approach with Local Search (LS) strategy to avoid local optima and then cloned into a binary version called Improved Binary AO (IBAO). Additionally, k-Nearest Neighbor (k-NN) and Support Vector Machine (SVM) are quality estimators. On 18 multi-scale benchmarks, the IBAO algorithm is compared with the original BAO algorithm and twelve recent algorithms, such as Binary Artificial Bee Colony (BABC), Binary Bat Algorithm (BBA), Binary Particle Swarm Optimization (BPSO), Binary Whale Optimization Algorithm (BWOA), Binary Grey Wolf Optimization (BGWO), Binary Sailfish Optimizer (BSFO), Binary Henry Gas Solubility Optimization (BHGSO), Binary Harris Hawks Optimization (BHHO). According to Wilcoxon’s rank-sum test (α=0.05), the dominance and significant influence of IBAO are apparent on both small and large dimensional benchmarks and obtained classification accuracy up to 100% in some of these benchmarks integrated with a feature reduction size down to 92%.

Adversarial Approaches to Tackle Imbalanced Data in Machine Learning

Real-world applications often involve imbalanced datasets, which have different distributions of examples across various classes. When building a system that requires a high accuracy, the performance of the classifiers is crucial. However, imbalanced datasets can lead to a poor classification performance and conventional techniques, such as synthetic minority oversampling technique. As a result, this study proposed a balance between the datasets using adversarial learning methods such as generative adversarial networks. The model evaluated the effect of data augmentation on both the balanced and imbalanced datasets. The study evaluated the classification performance on three different datasets and applied data augmentation techniques to generate the synthetic data for the minority class. Before the augmentation, a decision tree was applied to identify the classification accuracy of all three datasets. The obtained classification accuracies were 79.9%, 94.1%, and 72.6%. A decision tree was used to evaluate the performance of the data augmentation, and the results showed that the proposed model achieved an accuracy of 82.7%, 95.7%, and 76% on a highly imbalanced dataset. This study demonstrates the potential of using data augmentation to improve the classification performance in imbalanced datasets.

EEG-Based Emotion Classification in Financial Trading Using Deep Learning: Effects of Risk Control Measures

Day traders in the financial markets are under constant pressure to make rapid decisions and limit capital losses in response to fluctuating market prices. As such, their emotional state can greatly influence their decision-making, leading to suboptimal outcomes in volatile market conditions. Despite the use of risk control measures such as stop loss and limit orders, it is unclear if these strategies have a substantial impact on the emotional state of traders. In this paper, we aim to determine if the use of limit orders and stop loss has a significant impact on the emotional state of traders compared to when these risk control measures are not applied. The paper provides a technical framework for valence-arousal classification in financial trading using EEG data and deep learning algorithms. We conducted two experiments: the first experiment employed predetermined stop loss and limit orders to lock in profit and risk objectives, while the second experiment did not employ limit orders or stop losses. We also proposed a novel hybrid neural architecture that integrates a Conditional Random Field with a CNN-BiLSTM model and employs Bayesian Optimization to systematically determine the optimal hyperparameters. The best model in the framework obtained classification accuracies of 85.65% and 85.05% in the two experiments, outperforming previous studies. Results indicate that the emotions associated with Low Valence and High Arousal, such as fear and worry, were more prevalent in the second experiment. The emotions associated with High Valence and High Arousal, such as hope, were more prevalent in the first experiment employing limit orders and stop loss. In contrast, High Valence and Low Arousal (calmness) emotions were most prominent in the control group which did not engage in trading activities. Our results demonstrate the efficacy of our proposed framework for emotion classification in financial trading and aid in the risk-related decision-making abilities of day traders. Further, we present the limitations of the current work and directions for future research.

Classification of voltage sags causes in industrial power networks using multivariate time‐series

AbstractVoltage sags are the most frequent and impactful disturbances in industrial power grids, leading to high financial losses for industrial clients. The identification of the cause and its relative location is crucial for the contractual relation between the energy provider and the industrial customers. This paper proposes a methodology to identify the origins of voltage sags based on instantaneous symmetrical components and dynamic time warping. Short‐Time Fourier and Fortescue transform are implemented in the pre‐processing step using the voltage and current waveforms. Then, a distance‐based classification strategy to identify the sources of voltage sags is used. It relies on a four‐dimension time‐series signature used as features. Moreover, a confidence index associated with the classification output is provided. The proposal offers an easy implementation in industrial applications with no previous recorded data. It has the benefit of using a reduced‐size reference database entirely composed of synthetic data. The main advantages of the proposed method are its generalization capabilities and the possibility of raising an alert based on the confidence index. The obtained classification accuracy on synthetic data with seven causes is 100%. The method reaches a classification F1‐score higher than 99% with field measurements representing five classes obtained from three different industrial sites.

AdaD-FNN for Chest CT-Based COVID-19 Diagnosis

Coronavirus disease 2019 (COVID-19) generated a global public health emergency since December 2019, causing huge economic losses. To help radiologists strengthen their recognition of COVID-19 cases, we developed a computer-aided diagnosis system based on deep learning to automatically classify chest computed tomography-based COVID-19, Tuberculosis, and healthy control subjects. Our novel classification model AdaD-FNN sequentially transfers the trained knowledge of an FNN estimator to the next FNN estimator while updating the weights of the samples in the training set with a decaying learning rate. This model inhibits the network from remembering the noisy information and improves the learning of complex patterns in the hard-to-identify samples. Moreover, we designed a novel image preprocessing model F-U2MNet-C by enhancing the image features using fuzzy stacking and eliminating the interference factors using U2MNet segmentation. Extensive experiments are conducted on four publicly available datasets namely, TLDCA, UCSD-Al4H, SARS-CoV-2, TCIA, and the obtained classification accuracies are 99.52%, 92.96%, 97.86%, 91.97%. Our novel system gives out compelling performance for assisting COVID-19 detection when compared with 22 state-of-the-art methods. We hope to help link together biomedical research and artificial intelligence and to assist the diagnosis of doctors, radiologists, and inspectors at each epidemic prevention site in the real world.

Machine-Learning Methods for Material Identification Using mmWave Radar Sensor

In recent years, radar sensors are gaining a paramount role in noninvasive inspection of different objects and materials. In this article, we present a framework for using machine learning in material identification based on their reflected radar signature. We employ multiple receiving (RX) channels of the radar module to capture the signatures of the reflected signal from different target materials. Within the proposed framework, we present three approaches suitable for material classification, namely: 1) convolutional neural networks (CNNs); 2) <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${k}$ </tex-math></inline-formula> -nearest neighbor ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${k}$ </tex-math></inline-formula> -NN); and 3) dynamic time warping (DTW). The proposed framework is tested using extensive experimentation and found to provide near-ideal classification accuracy in classifying six distinct material types. Furthermore, we explore the possibility of utilizing the framework to detect the volume of the identified material, where the obtained classification accuracy is above 98% in distinguishing three different volume levels.

Detection of Parkinson’s disease from EEG signals using discrete wavelet transform, different entropy measures, and machine learning techniques

Early detection of Parkinson’s disease (PD) is very important in clinical diagnosis for preventing disease development. In this study, we present efficient discrete wavelet transform (DWT)-based methods for detecting PD from health control (HC) in two cases, namely, off-and on-medication. First, the EEG signals are preprocessed to remove major artifacts before being decomposed into several EEG sub-bands (approximate and details) using DWT. The features are then extracted from the wavelet packet-derived reconstructed signals using different entropy measures, namely, log energy entropy, Shannon entropy, threshold entropy, sure entropy, and norm entropy. Several machine learning techniques are investigated to classify the resulting PD/HC features. The effects of DWT coefficients and brain regions on classification accuracy are being investigated as well. Two public datasets are used to verify the proposed methods: the SanDiego dataset (31 subjects, 93 min) and the UNM dataset (54 subjects, 54 min). The results are promising and show that four entropy measures: log energy entropy, threshold entropy, sure entropy, and modified-Shannon entropy (TShEn) lead to high classification accuracy, indicating they are good biomarkers for PD detection. With the SanDiego dataset, the classification results of off-medication PD versus HC are 99.89, 99.87, and 99.91 for accuracy, sensitivity, and specificity, respectively, using the combination of DWT + TShEn and KNN classifier. Using the same combination, the results of on-medication PD versus HC are 94.21, 93.33, and 95%. With the UNM dataset, the obtained classification accuracy is around 99.5% in both cases of off-and on-medication PD using DWT + TShEn + SVM and DWT + ThEn + KNN, respectively. The results also demonstrate the importance of all DWT coefficients and that selecting a suitable small number of EEG channels from several brain regions could improve the classification accuracy.

An Optimal Approach for Heart Sound Classification Using Grid Search in Hyperparameter Optimization of Machine Learning.

Heart-sound auscultation is one of the most widely used approaches for detecting cardiovascular disorders. Diagnosing abnormalities of heart sound using a stethoscope depends on the physician's skill and judgment. Several studies have shown promising results in automatically detecting cardiovascular disorders based on heart-sound signals. However, the accuracy performance needs to be enhanced as automated heart-sound classification aids in the early detection and prevention of the dangerous effects of cardiovascular problems. In this study, an optimal heart-sound classification method based on machine learning technologies for cardiovascular disease prediction is performed. It consists of three steps: pre-processing that sets the 5 s duration of the PhysioNet Challenge 2016 and 2022 datasets, feature extraction using Mel frequency cepstrum coefficients (MFCC), and classification using grid search for hyperparameter tuning of several classifier algorithms including k-nearest neighbor (K-NN), random forest (RF), artificial neural network (ANN), and support vector machine (SVM). The five-fold cross-validation was used to evaluate the performance of the proposed method. The best model obtained classification accuracy of 95.78% and 76.31%, which was assessed using PhysioNet Challenge 2016 and 2022, respectively. The findings demonstrate that the suggested approach obtained excellent classification results using PhysioNet Challenge 2016 and showed promising results using PhysioNet Challenge 2022. Therefore, the proposed method has been potentially developed as an additional tool to facilitate the medical practitioner in diagnosing the abnormality of the heart sound.

A proposed approach for diabetes diagnosis using neuro-fuzzy technique

Diabetes is a chronic disease characterized by a decrease in pancreatic insulin production. The immune system will be harmed due to this condition, which will raise blood sugar levels. However, early detection of diabetes enables patients to begin treatment on time, therefore reducing or eliminating the risk of severe consequences. One of the most significant challenges in the healthcare unit is disease diagnosis. Traditional techniques of disease diagnosis are manual and prone to inaccuracy. This paper proposed an approach for diagnosing diabetes using the adaptive neuro-fuzzy inference system (ANFIS) based on Pima Indians diabetes dataset (PIDD). The three stages of the proposed approach are pre-processing classification and evaluation. Normalization, imputation, and anomaly detection are part of the pre-processing stage. The pre-processing was done by normalizing the data, replacing the missing values, and using the local outlier factor (LOF) technique. In the classification stage, ANFIS classifiers were trained using the hybrid learning algorithm of the neural network. Finally, the evaluation procedures use the last stage’s sensitivity, specificity, and accuracy metrics. The obtained classification accuracy was 92.77%, and it seemed rather promising compared to the other classification applications for this topic found in the literature.

Research on the Classification of ECG and PCG Signals Based on BiLSTM-GoogLeNet-DS

Because a cardiac function signal cannot reflect cardiac health in all directions, we propose a classification method using ECG and PCG signals based on BiLSTM-GoogLeNet-DS. The electrocardiogram (ECG) and phonocardiogram (PCG) signals used as research objects were collected synchronously. Firstly, the obtained ECG and PCG signals were filtered, and then the ECG and PCG signals were fused and classified by using a bi-directional long short-term memory network (BiLSTM). After that, the time-frequency processing was performed on the filtered ECG and PCG signals to obtain the time-frequency diagram of each signal; the one-dimensional signal was changed into a two-dimensional image signal, and the size of each image signal was adjusted to input the improved GoogLeNet network for classification. Then we obtained the two-channel classification results. The three-channel classification results, combined with the classification results of the above BiLSTM network, were finally obtained. The classification results of these three channels were used to make a decision via the fusion strategy of the improved D-S theory. Finally, we obtained the classification results. Taking 70% of the signals in the database as training data and 30% as test data, the obtained classification accuracy was 96.13%, the sensitivity was 98.48%, the specificity was 90.8%, and the F1 score was 97.24%. From the experimental results, the method proposed in this paper obtained high classification accuracy, and the classification effect was better than a cardiac function signal, which makes up for the low accuracy of the cardiac function signal for judging cardiac disease.

Diagnosis of Early Glottic Cancer Using Laryngeal Image and Voice Based on Ensemble Learning of Convolutional Neural Network Classifiers

The purpose of study is to improve the classification accuracy by comparing the results obtained by applying decision tree ensemble learning, which is one of the methods to increase the classification accuracy for a relatively small dataset, with the results obtained by the convolutional neural network (CNN) algorithm for the diagnosis of glottal cancer. Pusan National University Hospital (PNUH) dataset were used to establish classifiers and Pusan National University Yangsan Hospital (PNUYH) dataset were used to verify the classifier's performance in the generated model. For the diagnosis of glottic cancer, deep learning-based CNN models were established and classified using laryngeal image and voice data. Classification accuracy was obtained by performing decision tree ensemble learning using probability through CNN classification algorithm. In this process, the classification and regression tree (CART) method was used. Then, we compared the classification accuracy of decision tree ensemble learning with CNN individual classifiers by fusing the laryngeal image with the voice decision tree classifier. We obtained classification accuracy of 81.03 % and 99.18 % in the established laryngeal image and voice classification models using PNUH training dataset, respectively. However, the classification accuracy of CNN classifiers decreased to 73.88 % in voice and 68.92 % in laryngeal image when using an external dataset of PNUYH. To solve this problem, decision tree ensemble learning of laryngeal image and voice was used, and the classification accuracy was improved by integrating data of laryngeal image and voice of the same person. The classification accuracy was 87.88 % and 89.06 % for the individualized laryngeal image and voice decision tree model respectively, and the fusion of the laryngeal image and voice decision tree results represented a classification accuracy of 95.31 %. The results of our study suggest that decision tree ensemble learning aimed at training multiple classifiers is useful to obtain an increased classification accuracy despite a small dataset. Although a large data amount is essential for AI analysis, when an integrated approach is taken by combining various input data high diagnostic classification accuracy can be expected.