Maximum Classification Accuracy Research Articles

Selecting EEG channels and features using multi-objective optimization for accurate MCI detection: validation using leave-one-subject-out strategy

Effective management of dementia requires the timely detection of mild cognitive impairment (MCI). This paper introduces a multi-objective optimization approach for selecting EEG channels (and features) for the purpose of detecting MCI. Firstly, each EEG signal from each channel is decomposed into subbands using either variational mode decomposition (VMD) or discrete wavelet transform (DWT). A feature is then extracted from each subband using one of the following measures: standard deviation, interquartile range, band power, Teager energy, Katz's and Higuchi's fractal dimensions, Shannon entropy, sure entropy, or threshold entropy. Different machine learning techniques are used to classify the features of MCI cases from those of healthy controls. The classifier's performance is validated using leave-one-subject-out (LOSO) cross-validation (CV). The non-dominated sorting genetic algorithm (NSGA)-II is designed with the aim of minimizing the number of EEG channels (or features) and maximizing classification accuracy. The performance is evaluated using a publicly available online dataset containing EEGs from 19 channels recorded from 24 participants. The results demonstrate a significant improvement in performance when utilizing the NSGA-II algorithm. By selecting only a few appropriate EEG channels, the LOSO CV-based results show a significant improvement compared to using all 19 channels. Additionally, the outcomes indicate that accuracy can be further improved by selecting suitable features from different channels. For instance, by combining VMD and Teager energy, the SVM accuracy obtained using all channels is 74.24%. Interestingly, when only five channels are selected using NSGA-II, the accuracy increases to 91.56%. The accuracy is further improved to 95.28% when using only 8 features selected from 7 channels. This demonstrates that by choosing informative features or channels while excluding noisy or irrelevant information, the impact of noise is reduced, resulting in improved accuracy. These promising findings indicate that, with a limited number of channels and features, accurate diagnosis of MCI is achievable, which opens the door for its application in clinical practice.

Improving golden jackel optimization algorithm: An application of chemical data classification

One of the main issues affecting the effectiveness of the quantitative structure-activity relationship (QSAR) classification techniques in chemometrics is high dimensionality. Applying feature selection is a critical procedure that determines the most relevant and important aspects of a dataset. It improves the effectiveness and accuracy of prediction models by effectively lowering the number of features. This decrease increases classification accuracy, reduces computing strain, and improves overall performance. Recently, the golden jackal optimization (GJO) algorithm was introduced, which has been successfully used to solve various continuous optimization issues. Therefore, this study proposes an improvement in the GJO algorithm employing chaotic maps, abbreviated as CGJO, to enhance the exploration and exploitation capability of the GJO algorithm in picking the essential descriptors in QSAR classification models with high classification accuracy and less computation time. Experimental findings based on four different high-dimensional chemical datasets show that the proposed CGJO algorithm can maximize classification accuracy while simultaneously decreasing the number of chosen descriptors and lowering the time required for computing. Thus, the proposed algorithm can be useful for chemical data classification in other QSAR modeling.

Scene classification for remote sensing image of land use and land cover using dual-model architecture with multilevel feature fusion

ABSTRACT Scene classification for remote sensing image (RSI) of land use and land cover (LULC) involves identifying discriminative features of interest in different classes. Spurred by the powerful feature extraction capability of Convolutional Neural Networks (CNNs), LULC classification for RSI has rapidly developed in recent years. Although multi-models have better classification performance than single-models, combining multi-models is still the key to maximizing classification accuracy. Thus, this paper proposes a dual-model architecture with multilevel feature fusion called XE-Net. Specifically, high, middle, and low-level features are extracted by the Xception and EfficientNet-V2, respectively, and through transfer learning, the model’s weighting parameters from the ImageNet data are shared. Moreover, the designed sibling feature fusion algorithm fuses the triple-level features extracted from the dual models sequentially according to the same level. Besides, the proposed multi-scale feature fusion method systematically enhances the fused three-scale features to improve the discriminative feature. Finally, the discriminative feature is input into the classifier to obtain the classification results. The maximum average overall accuracy obtained from sufficient experiments using XE-Net on the RSSCN-7 dataset is 96.84%, while WHU-19, UCM-21, OPTIMAL-31, NWPU-RESISC45, and AID attain 99.58%, 99.37%, 97.07%, 95.03%, and 95.78%, respectively, demonstrating our model’s superiority.

A novel feature selection method based on quantum support vector machine

Feature selection is critical in machine learning to reduce dimensionality and improve model accuracy and efficiency. The exponential growth in feature space dimensionality for modern datasets directly results in ambiguous samples and redundant features, which can severely degrade classification accuracy. Quantum machine learning offers potential advantages for addressing this challenge. In this paper, we propose a novel method, quantum support vector machine feature selection (QSVMF), integrating quantum support vector machines with multi-objective genetic algorithm. QSVMF optimizes multiple simultaneous objectives: maximizing classification accuracy, minimizing selected features and quantum circuit costs, and reducing feature covariance. We apply QSVMF for feature selection on a breast cancer dataset, comparing the performance of QSVMF against classical approaches with the selected features. Experimental results show that QSVMF achieves superior performance. Furthermore, the Pareto front solutions of QSVMF enable analysis of accuracy versus feature set size trade-offs, identifying extremely sparse yet accurate feature subsets. We contextualize the biological relevance of the selected features in terms of known breast cancer biomarkers. This work highlights the potential of quantum-based feature selection to enhance machine learning efficiency and performance on complex real-world data.

Privacy-Preserving Breast Cancer Classification: A Federated Transfer Learning Approach.

Breast cancer is deadly cancer causing a considerable number of fatalities among women in worldwide. To enhance patient outcomes as well as survival rates, early and accurate detection is crucial. Machine learning techniques, particularly deep learning, have demonstrated impressive success in various image recognition tasks, including breast cancer classification. However, the reliance on large labeled datasets poses challenges in the medical domain due to privacy issues and data silos. This study proposes a novel transfer learning approach integrated into a federated learning framework to solve the limitations of limited labeled data and data privacy in collaborative healthcare settings. For breast cancer classification, the mammography and MRO images were gathered from three different medical centers. Federated learning, an emerging privacy-preserving paradigm, empowers multiple medical institutions to jointly train the global model while maintaining data decentralization. Our proposed methodology capitalizes on the power of pre-trained ResNet, a deep neural network architecture, as a feature extractor. By fine-tuning the higher layers of ResNet using breast cancer datasets from diverse medical centers, we enable the model to learn specialized features relevant to different domains while leveraging the comprehensive image representations acquired from large-scale datasets like ImageNet. To overcome domain shift challenges caused by variations in data distributions across medical centers, we introduce domain adversarial training. The model learns to minimize the domain discrepancy while maximizing classification accuracy, facilitating the acquisition of domain-invariant features. We conducted extensive experiments on diverse breast cancer datasets obtained from multiple medical centers. Comparative analysis was performed to evaluate the proposed approach against traditional standalone training and federated learning without domain adaptation. When compared with traditional models, our proposed model showed a classification accuracy of 98.8% and a computational time of 12.22s. The results showcase promising enhancements in classification accuracy and model generalization, underscoring the potential of our method in improving breast cancer classification performance while upholding data privacy in a federated healthcare environment.

A data driven approach to assess the petrophysical parametric sensitivity for lithology identification based on ensemble learning

Lithology identification is essential for geological analysis and reservoir characterization of hydrocarbon bearing formation. Conventional approaches of lithology identification are not only inconvenient but also time-consuming and expensive. Machine learning (ML) algorithms have been introduced as an effective, reliable, and accurate method of lithology identification using well-log data due to their ability to recognize the nonlinear relationship between log responses and lithology. In this study, we employed the XGBoost classifier model to assess the parametric sensitivity of the logging features in identifying lithology. Sequential Forward Feature Selection (SFFS) approach was applied to determine the most significant set of logging features that yields maximum classification accuracy. Nine logging parameters were used as the input of the XGBoost model to classify five lithology classes. Hyper-parameters were tuned within appropriate tuning search range to optimize the model performance along with Grid Search and 5-fold cross validation. The training and testing accuracy obtained from the model are 96% and 90.97%, respectively. The optimized XGBoost model showed the lowest precision, recall, and F1 score for the clay class. The shale class, on the other side, had the highest precision, recall, and F1 score of 95%, 94%, and 94%, respectively. The best set of logging features for identifying formation lithology is determined to be [PEF, SP, LLD, CALI, MSFL, NPHI, LLS], with a classification accuracy of 92.10% for the testing dataset. The least significant logging parameters for identifying lithology were found to be DT and GR, as their addition with the best input feature set, resulted in a negative deviation in classification accuracy. Finally, based on the classification accuracy, we ranked the top three sets of logging features, those contributed significantly.

Monitoring the condition of nitrogen-filled tires using weightless neural networks

The novelty of this paper revolves around monitoring the condition of nitrogen-filled tires through the fusion of features and utilization of weightless neural networks. A tire pressure monitoring system (TPMS) plays a crucial role in ensuring vehicle safety and comfort. Reckless driving, poor road conditions, continual operation, higher road friction and excessive load are certain factors that can degrade the longevity of tires. Such conditions can result in instantaneous fault attacks in tires raising a concern for safety and comfort. To apply instantaneous and accurate fault diagnosis, the present study leverages machine learning techniques through the integration of an adaptive and robust algorithm, namely, the Wilkes, Stonham and Aleksander Recognition Device (WiSARD) classifier. The experiment uses three types of features namely, statistical, histogram and autoregressive moving average (ARMA) features. The J48 decision tree algorithm was used to pinpoint the key attributes crucial for classification. Following this, the identified attributes were segregated into training and testing datasets, facilitating the evaluation of the WiSARD classifier. Hyperparameter tuning was carried out to achieve optimal value for maximizing classification accuracy and minimizing computational time. Among the features considered, ARMA features delivered the best test set accuracy of about 96.18%.

Accurate graph classification via two-staged contrastive curriculum learning.

Given a graph dataset, how can we generate meaningful graph representations that maximize classification accuracy? Learning representative graph embeddings is important for solving various real-world graph-based tasks. Graph contrastive learning aims to learn representations of graphs by capturing the relationship between the original graph and the augmented graph. However, previous contrastive learning methods neither capture semantic information within graphs nor consider both nodes and graphs while learning graph embeddings. We propose TAG (Two-staged contrAstive curriculum learning for Graphs), a two-staged contrastive learning method for graph classification. TAG learns graph representations in two levels: node-level and graph level, by exploiting six degree-based model-agnostic augmentation algorithms. Experiments show that TAG outperforms both unsupervised and supervised methods in classification accuracy, achieving up to 4.08% points and 4.76% points higher than the second-best unsupervised and supervised methods on average, respectively.

Improved salp swarm algorithm based on Newton interpolation and cosine opposition-based learning for feature selection

Feature selection (FS) is one of the most critical tasks in data mining, which aims to reduce the dimensionality of the data and maximize classification accuracy. The FS problem can be treated as an NP-hard problem. Recently, various swarm intelligent (SI) algorithms have been employed to deal with the FS problem to solve the expensive computation of the exact method. However, the performance of the SI algorithms is limited because these algorithms do not comprehensively take the characteristics of the FS problem into consideration. Therefore, a promising salp swarm algorithm called NCSSA is presented to solve this problem. In NCSSA, multi-perspective initialization strategy, Newton interpolation inertia weight, improved followers’ update model and cosine opposition-based learning (COBL) are proposed. In the majority of the SI algorithm-based FS method, the initial search agents are randomly generated or using a single filter method. However, a single filter method has different performance on various datasets. Therefore, a multi-perspective initialization strategy based on minimal redundancy maximal relevance (MRMR) and ReliefF is proposed, which can select the optimal subsets from different perspectives. Furthermore, Newton interpolation inertia weight is presented to balance the algorithm’s exploration and exploitation. Compare with the existing inertia weights, the adjustment flexibility of the proposed inertia weight is enhanced. Additionally, the followers update their positions according to the values of ReliefF and MRMR, which can make full use of the relationship between data and labels. Finally, the COBL is introduced to accelerate the convergence rate and helps the algorithm jump out of the local best solutions. The COBL is better than opposition-based learning (OBL) in terms of randomness, and considers the characteristics of the FS problem. The proposed NCSSA is compared to a series of non-SI-based methods and SI-based methods employing the standard datasets from the UCI Machine Learning Repository. Experimental results show that the NCSSA is a promising algorithm for the FS problem. The contribution analysis of each strategy indicates that the COBL is the most effective strategy in improving the SSA.

EVALUATION OF SPECTRAL INDICES FOR DETECTION OF BURNED AREAS IN AN ENVIRONMENTAL PROTECTION AREA USING SPOT-5 IMAGES

The Cerrado biome has great importance in ecological biodiversity, but deforestation has accelerated in recent decades. The Rio Preto-BA Environmental Preservation Area, known for the high agricultural activity of grains, continually suffers from forest fires and loss of native vegetation. Satellite remote sensing is proposed as an alternative to accurately locate and quantify fire-affected surfaces and their impacts on the landscape. With the recent free availability of SPOT-5 images, it has made it possible to detect burned areas with considerable spatial and spectral resolution. The objective of the present study was to determine the ideal spectral index for detection of the burned area inserted in the APA Rio Preto through two SPOT-5 scenes and the application of the Support Vector Machines (SVM) model. The burned areas detected were submitted to the calculation of five spectral indices, BAI, BAIM, NBR, NDVI and EVI2. The ability of each index to discriminate burned areas was estimated by comparing them with each other using a statistical separability index, SVM regression and accuracy analysis. BAIM was identified as the index with the greatest potential for discriminating burnt areas with maximum separability and classification accuracy above 90%, while the NDVI and EVI2 indices had low performance. It is hoped that these results can be used to evaluate and prioritize monitoring areas, contribute to the implementation of a fire management plan in the APA and to support subsequent studies on fire dynamics in forest systems integrated with advanced computational technologies.

Trace level detection of melamine and cyanuric acid extracted from pet liquid food (milk) using a SERS Au nanogap substrate

This study reported an application of Au nanogap substrates for surface-enhanced Raman scattering (SERS) measurements to quantitatively analyze melamine and its derivative products at trace levels in pet liquid food (milk) combined with a waveband selection approach, namely variable importance in projection (VIP). Six different concentrations of melamine, cyanuric acid, and melamine combined with cyanuric acid were created, and SERS spectra were acquired from 550 to 1620cm−1. Detection was possible up to 200 pM for melamine-contaminated samples, and 400 pM concentration detection for other two groups. The VIP-PLSR models obtained correlation coefficient (R2) values of 0.997, 0.985, and 0.981, with root mean square error of prediction (RMSEP) values of 18.492 pM, 19.777 pM, and 15.124 pM for prediction datasets. Additionally, partial least square discriminant analysis (PLS-DA) was used to classify both pure and different concentrations of spiked samples. The results showed that the maximum classification accuracy for melamine was 100%, for cyanuric acid it was 96%, and for melamine coupled with cyanuric acid it was 95%. The results obtained clearly demonstrated that the Au nanogap substrate offers low-concentration, rapid, and efficient detection of hazardous additive chemicals in pet consuming liquid food.

Design of computer algorithms based on parametric models

Abstract In this paper, the objective function is effectively optimized by improving the fitness function in the computer algorithm. The improvement mainly focuses on adjusting the weighting coefficients of completion time, load balance and execution cost. The article then proceeds to optimize the algorithm’s parameters based on the algorithm’s optimal parameterization criterion, and designs a hybrid hill-climbing-simulated annealing optimization algorithm based on the parameterized model. To verify the safety of the algorithm, avalanche effect experiments were conducted in the study. The experimental results show that when the number of iterations is 14, the number of changed bits reaches 32, indicating that the algorithm is susceptible to the avalanche effect. In addition, for the application effect of the algorithm, the study was tested in the Oliver30 dataset. When the number of iterations reaches 25000 times, the algorithm has reached the optimal value distance 432.82, which shows the superiority in urban path recognition performance. Meanwhile, the algorithm has achieved an accuracy 0.8286 for feature recognition in remote sensing datasets, with a maximum classification accuracy of more than 90%. The study provides critical practical data and theoretical support for further application of the computer algorithm.

Decreased wrist rotation imitation abilities in children with autism spectrum disorder.

While meaningless gross motor imitation (GMI) is a common challenge for children diagnosed with autism spectrum disorder (ASD), this topic has not attracted much attention and few appropriate test paradigms have been developed. The current study proposed a wrist rotation imitation (WRI) task (a meaningless GMI assignment), and established a WRI ability evaluation system using low-cost wearable inertial sensors, which acquired the simultaneous data of acceleration and angular acceleration during the WRI task. Three metrics (i.e., total rotation time, rotation amplitude, and symmetry) were extracted from those data of acceleration and angular acceleration, and then were adopted to construct classifiers based on five machine learning (ML) algorithms, including k-nearest neighbors, linear discriminant analysis, naive Bayes, support vector machines, and random forests. To illustrate our technique, this study recruited 49 ASD children (aged 3.5-6.5 years) and 59 age-matched typically developing (TD) children. Findings showed that compared with TD children, those with ASD may exhibit shorter total rotation time, lower rotation amplitude, and weaker symmetry. This implies that children with ASD might exhibit decreased WRI abilities. The classifier with the naive Bayes algorithm outperformed than other four algorithms, and achieved a maximal classification accuracy of 88% and a maximal AUC value of 0.91. Two metrics (i.e., rotation amplitude and symmetry) had high correlations with the gross and fine motor skills [evaluated by Gesell Developmental Schedules-Third Edition and Psychoeducational Profile-3 (PEP-3)]. While, the three metrics had no significant correlation with the visual-motor imitation abilities (evaluated by the subdomain of PEP-3) and the ASD symptom severity [evaluated by the Childhood Autism Rating Scale (CARS)] . The strengths of this study are associated with the low-cost measurement system, correlation between the WRI metrics and clinical measures, decreased WRI abilities in ASD, and high classification accuracy.

Universal screening in kindergarten: Validity and classification accuracy of Istation's Indicators of Progress-Early Reading.

Universal screening in reading is a common, and often required, practice in early elementary school. Computer-adaptive screening tools, such as Istation's Indicators of Progress-Early Reading (ISIP-ER), are often chosen for this purpose in schools. In our present study, we examine the validity evidence between the ISIP-ER in kindergarten and third grade State of Texas Assessments of Academic Readiness (STAAR) reading scores, the classification accuracy of ISIP-ER to predict which students will meet STAAR reading expectations, and a cut score to maximize classification accuracy for the local context. The sample included 962 students (Mage = 6.19 years; SDage = 0.37) from 15 elementary schools in one suburban school district in Texas. As for validity, the correlation between ISIP-ER in kindergarten and the third grade STAAR was moderate (r = 0.48). Classification accuracy analyses using the vendor-recommended cut score found sensitivity (0.63) and specificity (0.70) were all below recommended levels. Using a locally determined cut score, sensitivity (0.92) was improved, but specificity (0.33) was substantially decreased. The findings suggest ISIP-ER has some limitations in the accurate identification of students at risk for poor outcomes on a state-mandated reading test and will likely need to be combined with other assessments or progress monitoring data. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

An Integrated Active Deep Learning Approach for Image Classification from Unlabeled Data with Minimal Supervision

The integration of active learning (AL) and deep learning (DL) presents a promising avenue for enhancing the efficiency and performance of deep learning classifiers. This article introduces an approach that seamlessly integrates AL principles into the training process of DL models to build robust image classifiers. The proposed approach employs a unique methodology to select high-confidence unlabeled data points for immediate labeling, reducing the need for human annotation and minimizing annotation costs. Specifically, by combining uncertainty sampling with the pseudo-labeling of confident data, the proposed approach expands the training set efficiently. The proposed approach uses a hybrid active deep learning model that selects the most informative data points that need labeling based on an uncertainty measure. Then, it iteratively retrains a deep neural network classifier on the newly labeled samples. The model achieves high accuracy with fewer manually labeled samples than traditional supervised deep learning by selecting the most informative samples for labeling and retraining in a loop. Experiments on various image classification datasets demonstrate that the proposed model outperforms conventional approaches in terms of classification accuracy and reduced human annotation requirements. The proposed model achieved accuracy of 98.9% and 99.3% for the Cross-Age Celebrity and Caltech Image datasets compared to the conventional approach, which achieved 92.3% and 74.3%, respectively. In summary, this work presents a promising unified active deep learning approach to minimize the human effort in manually labeling data while maximizing classification accuracy by strategically labeling only the most valuable samples for the model.

Art appreciation model design based on improved PageRank and ECA-ResNeXt50 algorithm.

Image sentiment analysis technology can predict, measure and understand the emotional experience of human beings through images. Aiming at the problem of extracting emotional characteristics in art appreciation, this article puts forward an innovative method. Firstly, the PageRank algorithm is enhanced using tweet content similarity and time factors; secondly, the SE-ResNet network design is used to integrate Efficient Channel Attention (ECA) with the residual network structure, and ResNeXt50 is optimized to enhance the extraction of image sentiment features. Finally, the weight coefficients of overall emotions are dynamically adjusted to select a specific emotion incorporation strategy, resulting in effective bimodal fusion. The proposed model demonstrates exceptional performance in predicting sentiment labels, with maximum classification accuracy reaching 88.20%. The accuracy improvement of 21.34% compared to the traditional deep convolutional neural networks (DCNN) model attests to the effectiveness of this study. This research enriches images and texts' emotion feature extraction capabilities and improves the accuracy of emotion fusion classification.

Classification of lung pathologies in neonates using dual-tree complex wavelet transform

IntroductionUndiagnosed and untreated lung pathologies are among the leading causes of neonatal deaths in developing countries. Lung Ultrasound (LUS) has been widely accepted as a diagnostic tool for neonatal lung pathologies due to its affordability, portability, and safety. However, healthcare institutions in developing countries lack well-trained clinicians to interpret LUS images, which limits the use of LUS, especially in remote areas. An automated point-of-care tool that could screen and capture LUS morphologies associated with neonatal lung pathologies could aid in rapid and accurate diagnosis.MethodsWe propose a framework for classifying the six most common neonatal lung pathologies using spatially localized line and texture patterns extracted via 2D dual-tree complex wavelet transform (DTCWT). We acquired 1550 LUS images from 42 neonates with varying numbers of lung pathologies. Furthermore, we balanced our data set to avoid bias towards a pathology class.ResultsUsing DTCWT and clinical features as inputs to a linear discriminant analysis (LDA), our approach achieved a per-image cross-validated classification accuracy of 74.39% for the imbalanced data set. Our classification accuracy improved to 92.78% after balancing our data set. Moreover, our proposed framework achieved a maximum per-subject cross-validated classification accuracy of 64.97% with an imbalanced data set while using a balanced data set improves its classification accuracy up to 81.53%.ConclusionOur work could aid in automating the diagnosis of lung pathologies among neonates using LUS. Rapid and accurate diagnosis of lung pathologies could help to decrease neonatal deaths in healthcare institutions that lack well-trained clinicians, especially in developing countries.

An adaptive hybrid mutated differential evolution feature selection method for low and high-dimensional medical datasets

Feature selection (FS) constitutes a crucial endeavor in classification procedures, aiming to identify the minimal subset of features that maximizes classification accuracy. In the realm of combinatorial NP-hard challenges, FS assumes significance, prompting the utilization of robust metaheuristics as efficient wrapper-based FS strategies. However, the application of these wrapper metaheuristics to high-dimensional datasets, characterized by an abundance of features and a scarcity of samples, often results in a decline in effectiveness and escalated computational costs. Addressing these limitations, this study introduces the Adaptive Hybrid-Mutated Differential Evolution (A-HMDE) method, targeting the inherent drawbacks of the Differential Evolution (DE) algorithm. The A-HMDE incorporates four distinct strategies. Firstly, it integrates the mechanics of the Spider Wasp Optimization (SWO) algorithm into DE’s mutation strategies, yielding enhanced performance marked by high accuracy and swift convergence towards global optima. Secondly, adaptive mechanisms are applied to key DE parameters, amplifying the efficiency of the search process. Thirdly, an adaptive mutation operator ensures a harmonious balance between global exploration and local exploitation during optimization. Lastly, the concept of Enhanced Solution Quality (ESQ), rooted in the RUN algorithm, guides DE to elude local optima, thus heightening the accuracy of obtained solutions. The efficiency of the A-HMDE approach is assessed by employing it as a FS method across diverse datasets encompassing the UCI repository, microarray data, and skin disease image dataset. The experimental results accentuate the method’s remarkable ability to overcome challenges related to local minima and hasten the convergence process. A comprehensive comparison against contemporary cutting-edge algorithms highlights the considerable advancements achieved by the proposed method, with recorded accuracy ranging from 0.88 to 1.00.

Tree species classification on images from airborne mobile mapping using ML.NET

ABSTRACT Deep learning is a powerful tool for automating the process of recognizing and classifying objects in images. In this study, we used ML.NET, a popular open-source machine learning framework, to develop a model for identifying tree species in images obtained from airborne mobile mapping. These high-resolution images can be used to create detailed maps of the landscape. They can also be analyzed and processed to extract information about visual features, including tree species recognition. The deep learning model was trained using ML.NET to classify two tree species based on the combination of airborne mobile mapping images. Our approach yielded impressive results, with a maximum classification accuracy of 93.9%. This demonstrates the effectiveness of combining imagery sources with deep learning tools in ML.NET for efficient and accurate tree species classification. This study highlights the potential of the ML.NET framework for automating object classification and can provide valuable insights and information for forestry management and conservation efforts. The primary objective of this research was to evaluate the effectiveness of an approach for identifying tree species through a model generated using a combination of ortho and oblique images captured by a mobile mapping system.

Constraints on Optimising Encoder-Only Transformers for Modelling Sign Language with Human Pose Estimation Keypoint Data.

Supervised deep learning models can be optimised by applying regularisation techniques to reduce overfitting, which can prove difficult when fine tuning the associated hyperparameters. Not all hyperparameters are equal, and understanding the effect each hyperparameter and regularisation technique has on the performance of a given model is of paramount importance in research. We present the first comprehensive, large-scale ablation study for an encoder-only transformer to model sign language using the improved Word-level American Sign Language dataset (WLASL-alt) and human pose estimation keypoint data, with a view to put constraints on the potential to optimise the task. We measure the impact a range of model parameter regularisation and data augmentation techniques have on sign classification accuracy. We demonstrate that within the quoted uncertainties, other than ℓ2 parameter regularisation, none of the regularisation techniques we employ have an appreciable positive impact on performance, which we find to be in contradiction to results reported by other similar, albeit smaller scale, studies. We also demonstrate that the model architecture is bounded by the small dataset size for this task over finding an appropriate set of model parameter regularisation and common or basic dataset augmentation techniques. Furthermore, using the base model configuration, we report a new maximum top-1 classification accuracy of 84% on 100 signs, thereby improving on the previous benchmark result for this model architecture and dataset.