Bagged Tree Research Articles

Ensemble of bagged regression trees for concrete dam deformation predicting

Tree based ensemble learning algorithms involve generating multiple trees that together produce a consistent prediction and often significantly improving the accuracy of the prediction. In this paper, the technique is further explored, and a dam displacement prediction model based on bootstrap aggregating (bagged) regression trees (BART) is proposed. The case study of concrete dam indicates that the performance of bagged regression trees is better than the conventional regression tree (RT) model in predicting dam displacement. Moreover, the model can also analyze the importance of each variable, which has strong practical value.

Optimization of Irrelevant Features for Brain-Computer Interface (BCI) System

The brain is the most important body part for human. The brain controls all activities of the body such as movement, imagine, and response. Therefore, it is believed that the signals which collected from human scalp contain a lot of useful information. This useful information known as features can be extracted by applying advanced signal processing. Then, the features used for a brain-computer interface (BCI) system. However, the most suitable and relevant features for the BCI system still not investigate. In this paper, ten healthy subjects were involved in data collection. Threshold method, notch filter and wavelet decomposition were applied during pre-processing. Then, the signals were normalised. Hilbert-Huang Transform (HHT) and Power Spectral Density (PSD) were implemented. The features such as statistical-based features, approximate entropy (ApEn), sample entropy (SampEn), fuzzy entropy (FuzEn), permutation entropy (PermEn), distribution entropy (DistEn), Hjorth parameter, and Hurst exponent (HE) were extracted from PSD and HHT separately. Genetic algorithm (GA) and reliefF were carried out to select the most suitable and relevant features for the BCI system. The prediction rate before and after feature selection were compared. The performance after feature selection is improved in term of prediction rate and training time. The best classifier, in this case, is the bagged tree which can achieve 99.30%.

Video classification and retrieval through spatio-temporal Radon features

The rise in the availability of video content for access via the Internet and the medium of television has resulted in the development of automatic search procedures to retrieve the desired video. Searches can be simplified and hastened by employing automatic classification of videos. This paper proposes a descriptor called the Spatio-Temporal Histogram of Radon Projections (STHRP) for representing the temporal pattern of the contents of a video and demonstrates its application to video classification and retrieval. The first step in STHRP pattern computation is to represent any video as Three Orthogonal Planes (TOPs), i.e., XY, XT and YT, signifying the spatial and temporal contents. Frames corresponding to each plane are partitioned into overlapping blocks. Radon projections are obtained over these blocks at different orientations, resulting in weighted transform coefficients that are normalized and grouped into bins. Linear Discriminant Analysis (LDA) is performed over these coefficients of the TOPs to arrive at a compact description of STHRP pattern. Compared to existing classification and retrieval approaches, the proposed descriptor is highly robust to translation, rotation and illumination variations in videos. To evaluate the capabilities of the invariant STHRP pattern, we analyse the performance by conducting experiments on the UCF-101, HMDB51, 10contexts and TRECVID data sets for classification and retrieval using a bagged tree model. Experimental evaluation of video classification reveals that STHRP pattern can achieve classification rates of 96.15%, 71.7%, 93.24% and 97.3% for the UCF-101, HMDB51,10contexts and TRECVID 2005 data sets respectively. We conducted retrieval experiments on the TRECVID 2005, JHMDB and 10contexts data sets and the results revealed that STHRP pattern is able to provide the videos relevant to the user's query in minimal time (0.05s) with a good precision rate.

Toward a generalized predictive model of grapevine water status in Douro region from hyperspectral data

The predawn leaf water potential (ѱpd) is an eco-physiological indicator widely used for assessing vines water status and thus supporting irrigation management in several wine regions worldwide. However, the ѱpd is measured in a short time period before sunrise and the collection of a large sample of points is necessary to adequately represent a vineyard, which constitute operational constraints. In the present study, an alternative method based on hyperspectral data derived from a handheld spectroradiometer and machine learning algorithms was tested and validated for assessing grapevine water status. Two test sites in Douro wine region, integrating three grapevine cultivars, were studied for the years of 2014, 2015, and 2017. Four machine learning regression algorithms were tested for predicting the ѱpd as a continuous variable, namely Random Forest (RF), Bagging Trees (BT), Gaussian Process Regression (GPR), and Variational Heteroscedastic Gaussian Process Regression (VH-GPR). Three predicting variables, including two vegetation indices (NRI554,561 and WI900,970) and a time-dynamic variable based on the ѱpd (ѱpd_0), were applied for modelling the response variable (ѱpd). Additionally, the predicted values of ѱpd were aggregated into three classes representing different levels of water deficit (low, moderate, and high) and compared with the corresponding classes of ѱpd observed values. A root mean square error (RMSE) and a mean absolute error (MAE) lower or equal than 0.15 MPa and 0.12 MPa, respectively, were obtained with an external validation data set (n = 71 observations) for the various algorithms. When the modelling results were assessed through classes of values, a high overall accuracy was obtained for all the algorithms (82–83%), with prediction accuracy by class ranging between 79% and 100%. These results show a good performance of the predictive models, which considered a large variability of climatic, environmental, and agronomic conditions, and included various grape cultivars. By predicting both continuous values of ѱpd and classes of ѱpd, the approach presented in this study allowed obtaining 2-levels of accurate information about vines water status, which can be used to feed management decisions of different types of stakeholders.

Wind power prediction using bootstrap aggregating trees approach to enabling sustainable wind power integration in a smart grid

Precise prediction of wind power is important in sustainably integrating the wind power in a smart grid. The need for short-term predictions is increased with the increasing installed capacity. The main contribution of this work is adopting bagging ensembles of decision trees approach for wind power prediction. The choice of this regression approach is motivated by its ability to take advantage of many relatively weak single trees to reach a high prediction performance compared to single regressors. Moreover, it reduces the overall error and has the capacity to merge numerous models. The performance of bagged trees for predicting wind power has been compared to four commonly know prediction methods namely multivariate linear regression, support vector regression, principal component regression, and partial least squares regression. Real measurements recorded every ten minutes from an actual wind turbine are used to illustrate the prediction quality of the studied methods. Results showed that the bagged trees regression approach reached the highest prediction performance with a coefficient of determination of 0.982. The result showed that the bagged trees approach is followed by support vector regression with Gaussian kernel, the same model when using a quadratic kernel, and the multivariate linear regression, partial least squares, and principal component regression gave the lowest prediction. The investigated models in this study can represent a helpful tool for model-based anomaly detection in wind turbines.

Recognition of track defects through measured acceleration - part 2

For an optimized maintenance strategy, the early detection of track defects is necessary. Mounted sensors (e.g. acceleration sensors) on in-service trains are very suitable for track monitoring. With the continuous measurement of axle-box acceleration, short wavelength defects can be identified. For example, these defects can be rail breaks or cracks (i.e. rail defects), or local instabilities. Local instabilities can reduce the track quality in a short period of time. For an efficient data analysis of the acceleration signal and classification of different track defects, the development of appropriate methods is necessary. Therefore, a track-vehicle scale model was built to generate acceleration data for typical types of failures. With the generated dataset, developed algorithms for pattern recognition can be easily tested. In the second part of this research, three models created by the supervised learning method are trained and tested for the detection of the local instability in the vertical acceleration signal. The model A is trained with 78 laps and uses a manual classification. The chosen classifier for the model is a bagged tree algorithm implemented in the software MATLAB. The developed models distinguished between no failure, rail defect and local instability. For the training process of the model, the measured acceleration is treated statistically (e.g. Root-mean-square, Standard deviation, Spectral peaks and power). Subsequently test data for different scenarios is generated and used in the prediction model. With this model the track defects in the track-vehicle scale model are detected and classified very reliably. In contrast to existing methods, a machine learning approach is used for the non-destructive detection of the local instability. The results of the model are also improved by the model B and C by using 139 laps as training data, an automatic classification and an optimization of the statistics. The knowledge gained, can be used for acceleration data from in­service trains in regular operation, by adapting the developed model.

Proactive mental fatigue detection of traffic control operators using bagged trees and gaze-bin analysis

Most of existing eye movement-based fatigue detectors utilize statistical analysis of fixations, saccades, and blinks as inputs. Nevertheless, these parameters require long recording time and heavily depend on eye trackers. In an effort to facilitate proactive detection of mental fatigue, we introduced a complemental fatigue indicator, named gaze-bin analysis, which simply presents the eye-tracking data with histograms. A method which engaged the gaze-bin analysis as inputs of semisupervised bagged trees was developed. A case study in a vessel traffic service center demonstrated that this approach can alleviate the burden of manual labeling as well as improve the performance of fatigue detection model. In addition, the results show that the approach can achieve an excellent accuracy of 89%, which outperformed other methods. In general, this study provided a complemental indicator for detecting mental fatigue as well as enabled the application of a low sampling rate eye tracker in the traffic control center.

Energy load forecasting model based on deep neural networks for smart grids

In recent, smart grid has emerged as a promising technology to facilitate the future electric power grid and to balance between supply and demand. However, the intermittent nature of distributed energy resources causes dynamic uncertainties and nonlinearity in the smart grid environment. This may result in a large stress on power grid and has a big influence on energy planning, especially the generation and distribution. Therefore, energy load forecasting plays an important role in facilitating the operation of the future smart grid. Using the traditional statistical and machine learning approach there exists a significant forecasting error and high degree of overfitting. In this paper, we propose an energy load forecasting (ELF) model based on deep neural network architectures to manage the energy consumption in smart grids. First we investigate the applicability of two deep neural network architectures: deep feed-forward neural network (deep-FNN) and deep recurrent neural network (deep-RNN). To evaluate the models with low error, we simulate both architectures with multi size training set. Further, various activation functions and different combinations of hidden layer architectures are also tested. The simulation results are compared in terms of mean absolute percentage error. The results show that the proposed ELF model has attained better generalization and outperform the existing load forecasting models based on the shallow neural network, ensemble tree bagger and generalized linear regression.

Learning reciprocal actions for cooperative collision avoidance in quadrotor unmanned aerial vehicles

The ability to avoid collisions with each other is one of the fundamental requirements for autonomous unmanned aerial vehicles (UAVs) to be safely integrated into the civilian airspace, and for the viability of multi-UAV operations. This paper introduces a new approach for online cooperative collision avoidance between quadcopters, involving reciprocal maneuvers, i.e., coherent maneuvers without requiring any real-time consensus. Two maneuver strategies are presented, where UAVs respectively change their speed or heading to avoid a collision. A learning-based framework that trains these reciprocal actions for collision evasion (called TRACE) is developed. The primary elements of this framework include: 1) designing simulated experiments that cover a variety of UAV–UAV approach scenarios; 2) performing optimization to identify speed/heading change actions that satisfy safety constraints while minimizing the energy cost of the maneuver; and 3) using the offline optimization outcomes to train classifier (via ensemble bagged tree) and function approximation (via neural networks and Kriging) models for respectively selecting and encoding the avoidance actions. Trajectory generation and dynamics/controls are incorporated in the simulation environment used for training and testing. Over 90% accuracy in action prediction and over 95% success in avoiding collisions is observed when the trained models are applied to simulated unseen test scenarios.

Radiomic Biomarkers Evaluation of the High Dose Arm of RTOG 0617

The multi-institutional RTOG 0617 study compares the efficacy of high-dose (74 Gy) and standard-dose (60 Gy) radiotherapy for non-small cell lung cancer. The reported, unexpected outcome in the high-dose cohort not displaying a survival advantage differed from that observed in a previous multi-center study RTOG 0117. Patient heterogeneity may be attributed to the variation in radiotherapy effectiveness and might have explained why the high dose arm was not as effective as anticipated. This study aims to evaluate radiomic biomarkers with the potential to predict who would benefit from a high dose with predictive models built with radiomic characteristics. We analyzed 184 baseline planning CTs from DICOM images of patients in the high-dose cohort. The Radiomic features were extracted from the target structures within the GTV or ITV. Overall, 1130 radiomic characteristics were extracted from each patient using an open source software package. Furthermore, image pre-processing was performed with open source software. We used feature-selection methods, such as LASSO, Recursive Feature Elimination, forward selection, backward selection, Boruta, and Multivariate Adaptive Regression Spline (MARS), and evaluated their reduction efficiency and consistencies in feature selection. The MARS method provided the best results by identifying nine consistent features, which were used to build survival models. The accuracy of all models was estimated using 50-fold cross-validation. The Cox regression model was predictive with a C-index of 0.6552. The decision tree method exhibited the following: accuracy, 65.8%; AUC, 0.67; specificity, 70%; and sensitivity, 61%. The Ensemble Boosted Tree adopted an AdaBoost ensemble method with following characteristics: overall accuracy, 63%; AUC, 0.69; specificity, 73%; and sensitivity, 49%. The Ensemble Bagged Tree adopted the normal bootstrap aggregating method with the following characteristics: accuracy, 66.3%; AUC, 0.72; sensitivity, 77%; and specificity, 52%. Finally, the RUSBoosted tree method exhibited an accuracy of 65.8%, an AUC of 0.69, a sensitivity of 73%, and a specificity of 49%. Using the MARS method, we identified nine significant radiomic features. Using these features, 2-year survival predictive models were generated with reasonable performance. These nine features can also be used to calculate the prognostic index for patients to classify them into a high risk or low risk cohort. Hence determining who may benefit from a high dose prescription. Nevertheless, further investigation using more data and validation of the methodology is warranted.

Malicious Intrusion Detection Using Machine Learning Schemes

Wireless networks are continuously facing challenges in the field of Information Security. This leads to major researches in the area of Intrusion detection. The working of Intrusion detection is performed mainly by signature based detection and anomaly based detection. Anomaly based detection is based on the behavior of the network. One of the major challenge in this domain is to identify and detect the malicious node in wireless networks. The intrusion detection mechanism has to analyse the behavior of the node in the network by means of the several features possessed by each node. Intelligent schemes are the need of the hour in such scenario. This paper has taken a standard dataset for studying the features of the wireless node and reduced the features by applying the most efficient Correlation Attribute feature selection method. The machine learning algorithms are applied to obtain an effective training model which is then applied on the testing dataset to validate the model. The accuracy of the model is determined by the performance parameters such as true positive rate, false positive rate and ROC area. Neural network, bagging and decision tree algorithm RepTree are giving promising results in comparison with other classification algorithms.

A Fully Implantable Brain Machine Interface for Volitional Hand Grasp Restoration in Cervical Quadriplegia

Abstract INTRODUCTION Neural interface research has been strongly motivated by the need to restore communication and control to the estimated 5.3 million people in the US population which currently suffer from some form of paralysis. We recently enrolled a patient with a chronic traumatic cervical spinal cord injury (C5 ASIA A) to undergo placement of a brain machine interface aimed at restoring unilateral upper distal extremity function (ClinicalTrials.gov NCT02564419). The objective of this study was to evaluate the safety and efficacy of a fully implanted brain machine interface consisting of the Medtronic PC + S and the Bioness H200 hand rehabilitation orthosis for the functional restoration of hand grasp in a patient with cervical quadriplegia. METHODS A computer task was developed to engage the subject in thinking of either hand movement or rest while electrocorticographic (ECoG) activity was recorded. “Open-loop” trials were used to train various classifiers for predicting "move" or "rest" states based on observations of the ECoG activity. In "closed-loop" experiments, the decoded desired hand state was used to drive functional electrical stimulation of the dominant hand utilizing the Bioness H200 orthosis. Functional performance was measured by a modified Jebson Taylor Hand Function (JTHF) test and range of motion. RESULTS Movement intent information was decoded with an online accuracy of 88.2% with a tree bagging classifier over 21 sessions. Functional improvement was observed in reduction of the average time to perform subtasks within the JTHF test. CONCLUSION Our results demonstrate that a fully implanted brain machine interface can be safely implanted and used to reliably decode movement intent from motor cortex allowing for volitional control of hand grasp in a laboratory setting. Further work will aim to allow use of the device in a home setting, a critical step for the widespread use of these approaches to restore motor function in patients living with paralysis.

Tree-ring-width based streamflow reconstruction based on the random forest algorithm for the source region of the Yangtze River, China

Extending series of river streamflow based on tree-ring reconstruction is of scientific and practical importance for understanding hydrological or meteorological change of past. To achieve more accurate reconstructions, the intelligent learning algorithm random forest (RF) was proposed in this study to reconstruct the annual streamflow of the source region of the Yangtze River (SRYR). The method was developed using tree-ring chronologies ranging from 1485 to 2000 (AD) and annual streamflow from 1956 to 2000 (AD). The relationship between streamflow and the main large-scale atmospheric circulation as well as solar activity has also been discussed. The results show that: a) RF model could capture a more realistic characteristic of streamflow and show higher predictive ability for streamflow reconstruction than bagged regression trees (BRT), support vector machine (SVM), and simple linear regression (SLM). b) A period of lower streamflow occurred during the late 16th and mid-18th centuries, and the early 19th and mid-20th centuries experienced higher streamflow; an interesting temporal pattern indicated that the instrumental period was representative of individual highest (1979) and lowest (1989) streamflow years; in addition, a 2–8-year significant periodical oscillation (at 95% confidence level) was observed over most of the reconstructed series, with dominant periods of 2.5- and 4.9-year. c) The variability of streamflow in the study area was strongly associated with Pacific Decadal Oscillation (PDO), El Nino-Southern Oscillation (ENSO) and solar activity. This study provides reference for streamflow reconstruction based on tree-ring data and helps to understand the hydrological variation of past in SRYR.

Integration of Machine Learning and Open Access Geospatial Data for Land Cover Mapping

In-time and accurate monitoring of land cover and land use are essential tools for countries to achieve sustainable food production. However, many developing countries are struggling to efficiently monitor land resources due to the lack of financial support and limited access to adequate technology. This study aims at offering a solution to fill in such a gap in developing countries, by developing a land cover solution that is free of costs. A fully automated framework for land cover mapping was developed using 10-m resolution open access satellite images and machine learning (ML) techniques for the African country of Lesotho. Sentinel-2 satellite images were accessed through Google Earth Engine (GEE) for initial processing and feature extraction at a national level. Also, Food and Agriculture Organization’s land cover of Lesotho (FAO LCL) data were used to train a support vector machine (SVM) and bagged trees (BT) classifiers. SVM successfully classified urban and agricultural lands with 62 and 67% accuracy, respectively. Also, BT could classify the two categories with 81 and 65% accuracy, correspondingly. The trained models could provide precise LC maps in minutes or hours. they can also be utilized as a viable solution for developing countries as an alternative to traditional geographic information system (GIS) methods, which are often labor intensive, require acquisition of very high-resolution commercial satellite imagery, time consuming and call for high budgets.

Ensemble Bagged Tree Based Classification for Reducing Non-Technical Losses in Multan Electric Power Company of Pakistan

Non-technical losses (NTLs) have been a major concern for power distribution companies (PDCs). Billions of dollars are lost each year due to fraud in billing, metering, and illegal consumer activities. Various studies have explored different methodologies for efficiently identifying fraudster consumers. This study proposes a new approach for NTL detection in PDCs by using the ensemble bagged tree (EBT) algorithm. The bagged tree is an ensemble of many decision trees which considerably improves the classification performance of many individual decision trees by combining their predictions to reach a final decision. This approach relies on consumer energy usage data to identify any abnormality in consumption which could be associated with NTL behavior. The key motive of the current study is to provide assistance to the Multan Electric Power Company (MEPCO) in Punjab, Pakistan for its campaign against energy stealers. The model developed in this study generates the list of suspicious consumers with irregularities in consumption data to be further examined on-site. The accuracy of the EBT algorithm for NTL detection is found to be 93.1%, which is considerably higher compared to conventional techniques such as support vector machine (SVM), k-th nearest neighbor (KNN), decision trees (DT), and random forest (RF) algorithm.

IDHS-DSAMS: Identifying DNase I hypersensitive sites based on the dinucleotide property matrix and ensemble bagged tree

DNase I hypersensitive site (DHS) is related to DNA regulatory elements, so the understanding of DHS sites is of great significance for biomedical research. However, traditional experiments are not very good at identifying recombinant sites of a large number of emerging DNA sequences by sequencing. Some machine learning methods have been proposed to identify DHS, but most methods ignore spatial autocorrelation of the DNA sequence. In this paper, we proposed a predictor called iDHS-DSAMS to identify DHS based on the benchmark datasets. We develop a feature extraction method called dinucleotide-based spatial autocorrelation (DSA). Then we use Min-Redundancy-Max-Relevance (mRMR) to remove irrelevant and redundant features and a 100-dimensional feature vector is selected. Finally, we utilize ensemble bagged tree as classifier, which is based on the oversampled datasets using SMOTE. Five-fold cross validation tests on two benchmark datasets indicate that the proposed method outperforms its existing counterparts on the individual accuracy (Acc), Matthews correlation coefficient (MCC), sensitivity (Sn) and specificity (Sp).

Phenotyping Women Based on Dietary Macronutrients, Physical Activity, and Body Weight Using Machine Learning Tools.

Nutritional phenotyping can help achieve personalized nutrition, and machine learning tools may offer novel means to achieve phenotyping. The primary aim of this study was to use energy balance components, namely input (dietary energy intake and macronutrient composition) and output (physical activity) to predict energy stores (body weight) as a way to evaluate their ability to identify potential phenotypes based on these parameters. From the Women’s Health Initiative Observational Study (WHI OS), carbohydrates, proteins, fats, fibers, sugars, and physical activity variables, namely energy expended from mild, moderate, and vigorous intensity activity, were used to predict current body weight (both as body weight in kilograms and as a body mass index (BMI) category). Several machine learning tools were used for this prediction. Finally, cluster analysis was used to identify putative phenotypes. For the numerical predictions, the support vector machine (SVM), neural network, and k-nearest neighbor (kNN) algorithms performed modestly, with mean approximate errors (MAEs) of 6.70 kg, 6.98 kg, and 6.90 kg, respectively. For categorical prediction, SVM performed the best (54.5% accuracy), followed closely by the bagged tree ensemble and kNN algorithms. K-means cluster analysis improved prediction using numerical data, identified 10 clusters suggestive of phenotypes, with a minimum MAE of ~1.1 kg. A classifier was used to phenotype subjects into the identified clusters, with MAEs <5 kg for 15% of the test set (n = ~2000). This study highlights the challenges, limitations, and successes in using machine learning tools on self-reported data to identify determinants of energy balance.

Changing wind speed distributions under future global climate

Based on 12 global circulation models, the simulated changes in empirical 10 m wind speed distributions under the representative concentration pathway RCP8.5 were quantified. It is demonstrated that climate change leads to significant wind speed distribution changes around the world. It is very likely that mean wind speed increases in most parts of Brazil, South Africa and Eastern Australia whereas decreasing mean wind speed is projected for large parts of the USA and Eastern Russia. Changes in the other distribution moments (standard deviation, skewness and kurtosis) are simulated in almost every region of the world. The theoretical distributions Weibull, Kappa, Wakeby and Burr-Generalized Extreme Value were fitted to current and future empirical wind speed distributions. It was found that the simulated changes in distribution moments often lead to changes in the fitting accuracy. This fact highlights the great importance of considering the wind speed distribution as temporally highly variable. Based on these changes, the system of wind speed distributions (including the Burr-Generalized Extreme Value, Kappa and Wakeby distribution) was adapted to potential future climate conditions by optimizing a bagged trees classifier. The use of the system of wind speed distributions ensures a consistent wind speed distribution fitting by selecting the best fitting theoretical distribution under changing climate conditions. The results of this study are a valuable basis for planning future wind energy assessments from both the theoretical and practical perspective.

Comparing Deep Neural Networks, Ensemble Classifiers, and Support Vector Machine Algorithms for Object-Based Urban Land Use/Land Cover Classification

With the advent of high-spatial resolution (HSR) satellite imagery, urban land use/land cover (LULC) mapping has become one of the most popular applications in remote sensing. Due to the importance of context information (e.g., size/shape/texture) for classifying urban LULC features, Geographic Object-Based Image Analysis (GEOBIA) techniques are commonly employed for mapping urban areas. Regardless of adopting a pixel- or object-based framework, the selection of a suitable classifier is of critical importance for urban mapping. The popularity of deep learning (DL) (or deep neural networks (DNNs)) for image classification has recently skyrocketed, but it is still arguable if, or to what extent, DL methods can outperform other state-of-the art ensemble and/or Support Vector Machines (SVM) algorithms in the context of urban LULC classification using GEOBIA. In this study, we carried out an experimental comparison among different architectures of DNNs (i.e., regular deep multilayer perceptron (MLP), regular autoencoder (RAE), sparse, autoencoder (SAE), variational autoencoder (AE), convolutional neural networks (CNN)), common ensemble algorithms (Random Forests (RF), Bagging Trees (BT), Gradient Boosting Trees (GB), and Extreme Gradient Boosting (XGB)), and SVM to investigate their potential for urban mapping using a GEOBIA approach. We tested the classifiers on two RS images (with spatial resolutions of 30 cm and 50 cm). Based on our experiments, we drew three main conclusions: First, we found that the MLP model was the most accurate classifier. Second, unsupervised pretraining with the use of autoencoders led to no improvement in the classification result. In addition, the small difference in the classification accuracies of MLP from those of other models like SVM, GB, and XGB classifiers demonstrated that other state-of-the-art machine learning classifiers are still versatile enough to handle mapping of complex landscapes. Finally, the experiments showed that the integration of CNN and GEOBIA could not lead to more accurate results than the other classifiers applied.

Novel dynamic center based binary and ternary pattern network using M4 pooling for real world voice recognition

The signal processing is one the very important research area in the computer sciences and artificial intelligence. Because, audio recognition, voice activity detection, disease diagnosis, brain activity detection and predictions methods are evaluated using signal processing methods. Nowadays, deep methods have been become popular in the signal processing applications. In this article, a novel hybrid feature extraction network by using novel approximations and multiple pooling method. The proposed method uses both binary pattern (BP) and ternary pattern (TP) as feature extractor. In order to extract variable and distinctive features, dynamic center based feature extraction strategy is used. Hence, the proposed feature extraction network is called as dynamic center based binary and ternary pattern network (DC-BTPNet). The proposed DC-BTPNet is consists of 9 layer. Also, a novel multiple pooling method is used in DC-BTPNet. In order to select features, neighborhood component analysis (NCA) is utilized. Finally, the extracted features are forwarded to polynomial kernel support vector machine (SVM). In order to evaluate performance of the proposed method, a novel dataset is created. The proposed DC-BTPNet based multiple learning method achieved 89.0% accuracy rates and it was compared to other state-of-art convolutional networks. Other well-known conventional classifiers are also used for instance linear discriminant analysis (LDA), k nearest neighbor (KNN) and bagged tree (BT) classifiers are used to compare performance of the classifiers. The comparisons and results clearly proved success of the DC-BTPNet. These results demonstrated that the proposed methods can be achieved successful results in larger datasets.