Tree Species Discrimination in Tropical Forests Using Airborne Imaging Spectroscopy

The present study proposes a novel method to discriminate the sex and species of silkworm pupae using NIR spectroscopy (800–2778 nm). The spectra from 840 silkworm pupae were collected then divided into a calibration set (700) and a test set (140) using the Kennard–Stone (KS) algorithm. The recognition models were built using the radial basis function and neural network (RBF–NN) and support vector machine (SVM) approaches. The species and sex identification results using the RBF–NN and SVM models based on full spectral data achieved a low accuracy of 5% and 33.57%, respectively. To improve the accuracy and decrease the processing time, both principal component analysis (PCA) and linear discriminant analysis (LDA) were used to reduce the data dimensions. The performance of the optimized SVM model (92.14%) was much better than the RBF–NN model (19.29%) based on PCA. Overall, the best discrimination results were obtained using the RBF–NN and SVM models based on LDA, providing an accuracy of 100%. These promising results have shown that the LDA–SVM and LDA–RBF–NN models can accurately recognize the sex and species of silkworm pupae using NIR spectroscopy.

Abstract. Tree species mapping in tropical forests provides valuable insights for forest managers. Keystone species can be located for collection of seeds for forest restoration, reducing fieldwork costs. However, mapping of tree species in tropical forests using remote sensing data is a challenge due to high floristic and spectral diversity. Little is known about the use of different spectral regions as most of studies performed so far used visible/near-infrared (390-1000 nm) features. In this paper we show the contribution of shortwave infrared (SWIR, 1045-2395 nm) for tree species discrimination in a tropical semideciduous forest. Using high-resolution hyperspectral data we also simulated WorldView-3 (WV-3) multispectral bands for classification purposes. Three machine learning methods were tested to discriminate species at the pixel-level: Linear Discriminant Analysis (LDA), Support Vector Machines with Linear (L-SVM) and Radial Basis Function (RBF-SVM) kernels, and Random Forest (RF). Experiments were performed using all and selected features from the VNIR individually and combined with SWIR. Feature selection was applied to evaluate the effects of dimensionality reduction and identify potential wavelengths that may optimize species discrimination. Using VNIR hyperspectral bands, RBF-SVM achieved the highest average accuracy (77.4%). Inclusion of the SWIR increased accuracy to 85% with LDA. The same pattern was also observed when WV-3 simulated channels were used to classify the species. The VNIR bands provided and accuracy of 64.2% for LDA, which was increased to 79.8 % using the new SWIR bands that are operationally available in this platform. Results show that incorporating SWIR bands increased significantly average accuracy for both the hyperspectral data and WorldView-3 simulated bands.

We introduce a new idea of robust command recognition system using frequency domain analysis and radial basis function support vector machines (SVM). We have tested the proposed system under stationary background noise hypothesis and have compared the performance of SVM classifier to multi-layer perceptron and radial basis function neural network classifiers. Simulations were carried out on the TI-46 corpus and higher performance of the SVM classifier was obtained on a small set of commands.

As arriving at the Earth, Interplanetary Coronal Mass Ejections (ICME) will interact with the Earth's magnetosphere and cause geomagnetic storms. The ICME event set is obtained by Richardson and Cane's Near Earth ICME list, and the input features are extracted based on interplanetary solar wind and magnetic data during ICME disturbance. A total of 483 ICME events from 1996 to 2006 are chosen in this study. 13 magnetic and kinetic features are finally selected for the training of the machine learning model. Rank of each feature's Fisher score indicates that the duration of the south-directed interplanetary magnetic field that is larger than 10nT and the increase of solar wind speed at the upstream shock or wave disturbance is closely related to the geoeffectiveness of ICME events, which is consistent with those former statistical results. The trained Radial Basis Function Support Vector Machine (RBF-SVM) can determine whether an ICME event could trigger moderate or stronger geomagnetic storms (<i>Dst</i> ≤ -50nT) effectively with an accuracy of 0.78±0.08. The results show that RBF-SVM can be used as a powerful tool in further analysis, and the better prediction of the geoeffectiveness of ICME will be obtained.

Given the demand for developing the efficient Machine Learning (ML) classification models for healthcare data, and the potentiality of Bio-Inspired Optimization (BIO) algorithms to tackle the problem of high dimensional data, we investigate the range of ML classification models trained with the optimal subset of features of PD data set for efficient PD classification. We used two BIO algorithms, Genetic Algorithm (GA) and Binary Particle Swarm Optimization (BPSO), to determine the optimal subset of features of PD data set. The data set chosen for investigation comprises 756 observations (rows or records) taken over 755 attributes (columns or dimensions or features) from 252 PD patients. We employed MaxAbsolute feature scaling method to normalize the data and one hold cross-validation method to avoid biased results. Accordingly, the data is split in to training and testing set in the ratio of 70% and 30%. Subsequently, we employed GA and BPSO algorithms separately on 11 ML classifiers (Logistic Regression (LR), linear Support Vector Machine (lSVM), radial basis function Support Vector Machine (rSVM), Gaussian Naïve Bayes (GNB), Gaussian Process Classifier (GPC), k-Nearest Neighbor (kNN), Decision Tree (DT), Random Forest (RF), Multilayer Perceptron (MLP), Ada Boost (AB) and Quadratic Discriminant Analysis (QDA)), to determine the optimal subset of features (reduction of dimensionality) contributing to the highest classification accuracy. Among all the bio-inspired ML classifiers employed: GA-inspired MLP produced the maximum dimensionality reduction of 52.32% by selecting only 359 features and delivering 85.1% of the classification accuracy; GA-inspired AB delivered the maximum classification accuracy of 90.7% producing the dimensionality reduction of 41.43% by selecting only 441 features; And, BPSO-inspired GNB produced the maximum dimensionality reduction of 47.14% by selecting 396 features and delivering the classification accuracy of 79.3%; BPSOMLP delivered the maximum classification accuracy of 89% and produced 46.48% of the dimensionality reduction by selecting only 403 features.

Previous studies applying machine learning to predict severe respiratory failure in COVID-19 patients have shown inconsistent results due to variations in study populations and predictor variables. This study aimed to predict intensive care unit admission and identify key predictive factors. This retrospective cohort study included patients with COVID-19 who presented with mild respiratory failure, most of whom received oxygen via a mask or nasal cannula. Eight machine learning algorithms-XGBoost, support vector machines, neural networks, k-nearest neighbors, random forest, decision trees, logistic regression, and naïve Bayes-were applied to predict intensive care unit admission. A total of 392 patients (63.5% male, mean age, 55.0 ± 15.3 years) were included in the study. During follow-up, 80 patients (20.4%) required intensive care unit admission. Among them, 320 (81.6%) received steroid therapy, 301 (76.8%) underwent pulse steroid therapy, and 76 (19%) had been vaccinated. The multilayer perceptron, XGBoost, and radial basis function support vector machine models achieved the best overall performance based on ROC-AUC and accuracy values (ROC-AUC: 0.75, 0.70, and 0.71; accuracy: 0.79, 0.79, and 0.79, respectively). The strongest predictors of intensive care unit admission were low lymphocyte count on the first day, as well as high age, ferritin, body mass index, Charlson Comorbidity Index, and computed tomography score. Machine learning algorithms can reliably predict intensive care unit admission in COVID-19 patients with mild respiratory failure. These models identified key clinical and laboratory factors that may facilitate early risk stratification and guide treatment planning.

Given a protein that is localized in the mitochondria it is very important to know the submitochondria localization of that protein to understand its function. In this work, we propose a submitochondria localizer whose feature extraction method is based on the Chou's pseudo-amino acid composition. The pseudo-amino acid based features are obtained by combining pseudo-amino acid compositions with hundreds of amino-acid indices and amino-acid substitution matrices, then from this huge set of features a small set of 15 "artificial" features is created. The feature creation is performed by genetic programming combining one or more "original" features by means of some mathematical operators. Finally, the set of combined features are used to train a radial basis function support vector machine. This method is named GP-Loc. Moreover, we also propose a very few parameterized method, named ALL-Loc, where all the "original" features are used to train a linear support vector machine. The overall prediction accuracy obtained by GP-Loc is 89% when the jackknife cross-validation is used, this result outperforms the performance obtained in the literature (85.2%) using the same dataset. While the overall prediction accuracy obtained by ALL-Loc is 83.9%.

In this paper, a new approach for pedestrian detection is presented. We design an ensemble of classifiers that employ different feature representation schemes of the pedestrian images: Laplacian EigenMaps, Gabor filters, and invariant local binary patterns. Each ensemble is obtained by varying the patterns used to train the classifiers and extracting from each image two feature vectors for each feature extraction method: one for the upper part of the image and one for the lower part of the image. A different radial basis function support vector machine (SVM) classifier is trained using each feature vector; finally, these classifiers are combined by the ldquosum rule.rdquo Experiments are performed on a large data set consisting of 4000 pedestrian and more than 25 000 nonpedestrian images captured in outdoor urban environments. Experimental results confirm that the different feature representations give complementary information, which has been exploited by fusion rules, and we have shown that our method outperforms the state-of-the-art approaches among pedestrian detectors.

Flooding causes several threats with outcomes which include peril to human and animal life, damage to property, and adversity to agricultural fields. Therefore, flood prediction is of prime importance for reducing loss of life and devastation to property. To model complex nature of hydrologic processes artificial neural network (ANN) tool is effectively being utilized for modelling different nonlinear relationships, and has proved to be an appropriate method for flood prediction. Present study investigated relative accuracy of radial basis function neural network (RBFNN) and support vector machine (SVM) models combined with Firefly Algorithm (FA) in predicting river flood discharge and contrasted with that of regular ANN, RBFNN and SVM models. Monthly river flow data of Silchar and Dholai stations located in Cachar district of Assam, India are utilized for the present study. For assessing model performance, coefficient of determination (R2), mean square error (MSE) and root mean square error (RMSE) were measured. Evaluation of outcomes shows that both RBF-FA (radial basis function - firefly algorithm) and SVM-FA (support vector machine - firefly algorithm) hybrid models give more precise forecasting results than RBFNN, FFBPNN (feed forward back propagation neural network) and SVM models. Yet, it can be observed that SVM-FA model give better forecasting outputs with R2 value 0.9818 than RBF-FA model. Results also reveal that simple SVM model performs marginally better than simple ANN model.

Linear and Quadratic Discriminant Analysis have been used widely in many areas of data mining, machine learning, and bioinformatics. Friedman proposed a compromise between Linear and Quadratic Discriminant Analysis, called Regularized Discriminant Analysis (RDA), which has been shown to be more flexible in dealing with various class distributions. RDA applies the regularization techniques by employing two regularization parameters, which are chosen to jointly maximize the classification performance. The optimal pair of parameters is commonly estimated via cross-validation from a set of candidate pairs. It is computationally prohibitive for high dimensional data, especially when the candidate set is large, which limits the applications of RDA to low dimensional data.In this paper, a novel algorithm for RDA is presented for high dimensional data. It can estimate the optimal regularization parameters from a large set of parameter candidates efficiently. Experiments on a variety of datasets confirm the claimed theoretical estimate of the efficiency, and also show that, for a properly chosen pair of regularization parameters, RDA performs favorably in classification, in comparison with other existing classification methods.

Sex estimation is an important part of skeletal analysis and forensic identification. Traditionally pelvic traits are utilized for accurate sex estimation. However, the long bones, especially humerus, have been proved to be as effective for determine the sex of the individual.The aim of this study was to compare the predictive accuracy of seven statistical modelling techniques including classical statistical methods and machine learning algorithms, to assess the sexual dimorphism of humerus on a French sample based on a metric analysis of 26 measurements. A total of 98 humeral bones (divided in two samples) were measured. Seven statistical models were compared: Linear Discriminant Analysis (LDA), Regularized Discriminant Analysis (RDA), Penalized Logistic Regression (PLR), Flexible Discriminant Analysis (FDA), Support Vector Machine (SVM), and Artificial Neural Network (ANN) and Random Forest (RF).With cross validation, classification accuracy was greater than 90% (ranges between 92% and 98%) for all models without variable selection methods. The simplification of the models has improved the accuracy between 98% and 100% and also a reduction of the number of variables to 6 or less. Penalized logistic regression (PLR), Random Forest (RF) and Linear discriminant analysis (LDA) were the best accuracy models.The measurements made at the proximal part of the humerus (WTT, CSD), at distal part (BEW, WT, MAW, THT) and of the entire bone (PLCT) stand out among the various models.The present study suggests that the humerus is an interesting alternative for sex estimation and that non-classical statistical models can provide a new approach.

The anti-programmed cell death protein-1 (PD-1) immune checkpoint inhibitor nivolumab is currently approved for the treatment of patients with metastatic renal cell carcinoma (mRCC); approximately 25% of patients respond. We hypothesized that we could identify a biomarker of response using radiomics to train a machine learning classifier to predict nivolumab response outcomes. Patients with mRCC of different histologies treated with nivolumab in a single institution between 2013 and 2017 were retrospectively identified. Patients were labelled as responders (complete response [CR]/partial response [PR]/durable stable disease [SD]) or non-responders based on investigator tumor assessment using RECIST 1.1 criteria. For each patient, lesions were contoured from pre-treatment and first post-treatment computed tomography (CT) scans. This information was used to train a radial basis function support vector machine classifier to learn a prediction rule to distinguish responders from non-responders. The classifier was internally validated by a 10-fold nested cross-validation. Thirty-seven patients were identified; 27 (73%) met the inclusion criteria. One hundred and four lesions were contoured from these 27 patients. The median patient age was 56 years, 78% were male, 89% had clear-cell histology, 89% had prior nephrectomy, and 89% had prior systemic therapy. There were 19 responders vs. eight non-responders. The lesions selected were lymph nodes (60%), lung metastases (23%), and renal/adrenal metastases (17%). For the classifier trained on the baseline CT scans, 69% accuracy was achieved. For the classifier trained on the first post-treatment CT scans, 66% accuracy was achieved. The set of radiomic signatures was found to have limited ability to discriminate nivolumab responders from non-responders. The use of novel texture features (two-point correlation measure, two-point cluster measure, and minimum spanning tree measure) did not improve performance.

Comparison of the performance of multiclass classifiers in chemical data: Addressing the problem of overfitting with the permutation test

Tropical forests have exceptional floristic diversity, but their characterization remains incomplete, in part due to the resource intensity of in-situ assessments. Remote sensing technologies can provide valuable, cost-effective, large-scale insights. This study investigates the combined use of airborne LiDAR and imaging spectroscopy to map tree species at landscape scale in French Guiana. Binary classifiers were developed for each of 20 species using linear discriminant analysis (LDA), regularized discriminant analysis (RDA) and logistic regression (LR). Complementing visible and near infrared (VNIR) spectral bands with short wave infrared (SWIR) bands improved the mean average classification accuracy of the target species from 56.1% to 79.6%. Increasing the number of non-focal species decreased the success rate of target species identification. Classification performance was not significantly affected by impurity rates (confusion between assigned classes) in the non-focal class (up to 5% of bias), provided that an adequate criterion was used for adjusting threshold probability assignment. A limited number of crowns (30 crowns) in each species class was sufficient to retrieve correct labels effectively. Overall canopy area of target species was strongly correlated to their basal area over 118 ha at 1.5 ha resolution, indicating that operational application of the method is a realistic prospect (R2 = 0.75 for six major commercial tree species).

With the speedy prosperity of artificial intelligence and sensor technology, the application of action recognition in football is becoming increasingly widespread. However, due to the rapid changes, complex dynamics, and diverse characteristics of football movements, traditional recognition methods face significant challenges in terms of real-time performance and accuracy. Based on this background, a football recognition model combining network wearable sensors and improved support vector machine classification algorithm is proposed. Firstly, by integrating the attitude data of accelerometers and gyroscopes, real-time dynamic features such as pitch angle, roll angle, and yaw angle are calculated, and principal component analysis is used for dimensionality reduction processing. Subsequently, the support vector machine model is optimized based on Gaussian kernel function and dynamic weighting strategy to improve classification accuracy and stability. Finally, a football action recognition model is constructed by combining wearable sensors with an improved support vector machine classification algorithm. The experiment outcomes show that the improved support vector machine algorithm achieves an action recognition accuracy of 93.8%, a recall rate of 92.5%, an F1 value of 0.92, and an inference time of 10.2 ms, which are significantly better than the comparative algorithm. In practical applications, the recognition model built has an accuracy rate of over 90% in recognizing four types of actions: standing, running, passing, and shooting, with an average recognition time as low as 9.4 ms. The research provides an efficient solution for intelligent football action recognition technology and lays the foundation for the practical application of multi-modal data fusion.