Improve Classification Accuracy Research Articles

Land Cover Classification in Bamori Range using Machine Learning Algorithms: A Comparative Study

Accurate land cover classification is essential for monitoring the environment and managing natural resources sustainably. In this study, we focused on the Bamori Range in the Guna district of Madhya Pradesh, India, using highresolution satellite imagery and machine learning to classify land into five categories: bare land, agriculture, fallow cropland, dense forest, and forest. To achieve this, we created a detailed dataset using Sentinel-2 imagery and Dynamic World probabilities, along with feature engineering to improve classification accuracy. We then tested the performance of several models, including Random Forest, Neural Networks, Enhanced Neural Networks, and a hyperparameter-tuned Random Forest, to see which worked best for this task

PCB Defect Classification with Data Augmentation-Based Ensemble Method for Sustainable Smart Manufacturing

In the rapidly evolving field of printed circuit board (PCB) manufacturing, automated optical inspection (AOI) systems play a critical role but often face challenges such as computational inefficiencies, high costs, and limited defect data. To address these issues, we propose an ensemble methodology that combines lightweight models with custom data augmentation techniques to enhance defect classification accuracy in real-time production environments. Our approach mitigates overfitting in small datasets by generating diverse models through advanced data augmentation and employing feature-specific validation strategies. These models are integrated into an ensemble framework, achieving complementary results that improve classification accuracy while reducing computational overhead. We validate the proposed method using two datasets: the general classification dataset CIFAR-10 and an on-site real-world PCB dataset. With our approach, the average accuracy on CIFAR-10 improved from 97.6% to 98.2%, and the accuracy on the PCB dataset increased from 81% to 89%. These results demonstrate the method’s effectiveness in addressing data scarcity and computational challenges in real-world manufacturing scenarios. By improving quality control and reducing waste, our method optimizes production processes and contributes to sustainability through cost savings and environmental benefits. The proposed methodology is versatile, scalable, and applicable to a range of defect classification tasks beyond PCB manufacturing, making it a robust solution for modern production systems.

Landscape Character Classification with a Deep Neural Network: A Case Study of the Jianghan Plain

Grounded in the theoretical and methodological frameworks of landscape character identification from the European Landscape Map (LANMAP) and landscape character assessment (LCA), this study developed an AI-based tool for landscape character analysis to classify the Jianghan Plain’s landscape more effectively. The proposed method leveraged a deep learning model, the artificial intelligence-based landscape character (AI-LC) classifier, along with specific naming and coding rules for the unique landscape character of the Jianghan Plain. Experimental results showed a significant improvement in classification accuracy, reaching 89% and 86% compared to traditional methods. The classifier identified 10 macro-level and 18 meso-level landscape character types within the region, which were further categorized into four primary zones—a lake network river basin, a hillfront terrace, surrounding mountains, and a lake network island hill—based on natural and social features. These advancements contributed to the theoretical framework of landscape character assessment, offering practical insights for landscape planning and conservation while highlighting AI’s transformative potential in environmental research and management.

Brain Tumour Detection Using Deep Learning and Image Processing

Abstract —Brain tumor detection is a critical aspect of neurooncology, where timely and accurate diagnosis significantly impacts patient outcomes. Traditional imaging techniques, while useful, are often limited by variations in image quality, illumination inconsistencies, and the complex nature of brain tumors themselves, which vary widely in size, shape, and texture. This paper presents a deep learning- based approach to brain tumor detection using Convolutional Neural Networks (CNNs) combined with advanced image processing methods. Our methodology involves preprocessing brain MRI scans using histogram equalization, morphological operations, and data augmentation to address illumination issues and enhance the tumor region. Following preprocessing, the images are fed into a CNN, where transfer learning from the Sequential model is applied to improve classification accuracy. Experimental results demonstrate high recall and precision rates, affirming the model’s robustness in detecting brain tumors in MRI images. This approach not only addresses the challenge of inconsistent MRI scan quality but also reduces the risk of misclassification, showing promise for clinical application. Keywords— Brain Tumor Detection, Computer-aided Diagnosis, Computer Vision, Convolutional Neural Networks, Deep Learning, Image Processing, Transfer Learning.

Enhanced brain tumor diagnosis using combined deep learning models and weight selection technique.

Brain tumor classification is a critical task in medical imaging, as accurate diagnosis directly influences treatment planning and patient outcomes. Traditional methods often fall short in achieving the required precision due to the complex and heterogeneous nature of brain tumors. In this study, we propose an innovative approach to brain tumor multi-classification by leveraging an ensemble learning method that combines advanced deep learning models with an optimal weighting strategy. Our methodology integrates Vision Transformers (ViT) and EfficientNet-V2 models, both renowned for their powerful feature extraction capabilities in medical imaging. This model enhances the feature extraction step by capturing both global and local features, thanks to the combination of different deep learning models with the ViT model. These models are then combined using a weighted ensemble approach, where each model's prediction is assigned a weight. To optimize these weights, we employ a genetic algorithm, which iteratively selects the best weight combinations to maximize classification accuracy. We trained and validated our ensemble model using a well-curated dataset comprising labeled brain MRI images. The model's performance was benchmarked against standalone ViT and EfficientNet-V2 models, as well as other traditional classifiers. The ensemble approach achieved a notable improvement in classification accuracy, precision, recall, and F1-score compared to individual models. Specifically, our model attained an accuracy rate of 95%, significantly outperforming existing methods. This study underscores the potential of combining advanced deep learning models with a genetic algorithm-optimized weighting strategy to tackle complex medical classification tasks. The enhanced diagnostic precision offered by our ensemble model can lead to better-informed clinical decisions, ultimately improving patient outcomes. Furthermore, our approach can be generalized to other medical imaging classification problems, paving the way for broader applications of AI in healthcare. This advancement in brain tumor classification contributes valuable insights to the field of medical AI, supporting the ongoing efforts to integrate advanced computational tools in clinical practice.

Lesiondiff: enhanced breast cancer classification via dynamic lesion amplification using diffusion models

ABSTRACT Breast cancer is a leading cause of mortality among women, underscoring the critical need for accurate and early diagnosis to enhance treatment efficacy. Traditional imaging techniques are limited in their ability to differentiate between benign and malignant lesions, particularly in the early stages, for there are very few images available for the lesion area and the resolution of these images is poor. This paper introduces a novel lesion diffusion model that dynamically amplifies lesion areas, providing a multi-frame analysis to improve classification accuracy. By integrating time-aware motion modeling, the proposed method tracks temporal changes in lesions, Generating a sequence of magnified frames highlighting subtle lesion features. Tested on the BUSI breast ultrasound dataset, our model achieved a 10.269% improvement in classification accuracy over baseline methods, with an average gain of 4.645% across multiple frames. The results demonstrate the model’s ability to enhance the claim and diagnostic utility of breast cancer images after magnification This dynamic lesion amplification approach presents a significant advancement in computer-aided breast cancer diagnostics, offering new possibilities for improving early-stage detection.

Review on Application of Chi-square Statistic in Text Classification in Recent Five Years

Abstract. The swift expansion of online textual data has rendered text classification increasingly vital in information management. Despite the prevalent usage of the chi-square test in text classification, there has been a scarcity of thorough research regarding its specific uses in recent years. Therefore, it is vital to encapsulate the research about the use of the chi-square test in text classification throughout the last five years. This report reviews the application of the chi-square statistic in Arabic text classification, social media data analysis, and medical literature classification and analyses its effectiveness in feature selection and enhancing classification performance. By reviewing and analyzing the academic literature, this report summarizes the application of improved chi-square feature selection methods to different text data types. It explores the effectiveness of these methods in improving classification accuracy. The findings indicate that chi-square has significant advantages in text classification in different domains, especially when dealing with complex linguistic texts and user-generated content.

Efficient and Accurate Brain Tumor Classification Using Hybrid MobileNetV2-Support Vector Machine for Magnetic Resonance Imaging Diagnostics in Neoplasms.

Magnetic Resonance Imaging (MRI) plays a vital role in brain tumor diagnosis by providing clear visualization of soft tissues without the use of ionizing radiation. Given the increasing incidence of brain tumors, there is an urgent need for reliable diagnostic tools, as misdiagnoses can lead to harmful treatment decisions and poor outcomes. While machine learning has significantly advanced medical diagnostics, achieving both high accuracy and computational efficiency remains a critical challenge. This study proposes a hybrid model that integrates MobileNetV2 for feature extraction with a Support Vector Machine (SVM) classifier for the classification of brain tumors. The model was trained and validated using the Kaggle MRI brain tumor dataset, which includes 7023 images categorized into four types: glioma, meningioma, pituitary tumor, and no tumor. MobileNetV2's efficient architecture was leveraged for feature extraction, and SVM was used to enhance classification accuracy. The proposed hybrid model showed excellent results, achieving Area Under the Curve (AUC) scores of 0.99 for glioma, 0.97 for meningioma, and 1.0 for both pituitary tumors and the no tumor class. These findings highlight that the MobileNetV2-SVM hybrid not only improves classification accuracy but also reduces computational overhead, making it suitable for broader clinical use. The MobileNetV2-SVM hybrid model demonstrates substantial potential for enhancing brain tumor diagnostics by offering a balance of precision and computational efficiency. Its ability to maintain high accuracy while operating efficiently could lead to better outcomes in medical practice, particularly in resource limited settings.

Ensemble Voting Method for Phonocardiogram Heart Signal Classification Using FFT Features

Heart disease is still one of the leading causes of death worldwide, hence the need for effective diagnostic tools. Phonocardiogram (PCG) signals have been explored as a complementary approach to electrocardiogram (ECG) to detect cardiac abnormalities. This research investigates the classification of PCG signals using Fast Fourier Transform (FFT) features and deep learning models, including Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and Temporal Convolutional Network (TCN). Hyperparameter tuning, particularly learning rate adjustment, is applied to optimize the performance of the models. The results show that the GRU and TCN models outperform the LSTM, achieving up to 92% accuracy at a learning rate of 0.0001. Ensemble learning with soft voting was also applied to combine the strengths of each model. Although the ensemble model showed strong performance with 92% accuracy and ROC AUC of 0.9636, it did not provide significant improvement over the base model. This finding highlights the importance of hyperparameter tuning in model optimization, with GRU and TCN showing slightly better performance in the time series classification task. This study concludes that ensemble learning offers stability but does not significantly improve classification accuracy beyond a well-tuned base model.

Head Tumor Segmentation and Detection Based on Resunet

Deep learning algorithms are designed to mimic the functions of the human brain. These algorithms can simulate the human brain for feature extraction, i.e. neural networks with many hidden layers. This algorithm is being trained and tested using the MRI brain tumor segmentation dataset provided by Kaggle. ResNet50 is used for tumor classification tasks due to its powerful feature extraction capability. Through its deep residual structure, ResNet50 can effectively extract different lesion features and improve classification accuracy. At the same time, ResUNet combines the feature extraction capabilities of ResNet with the segmentation advantages of U-Net, accurately capturing tumor boundaries and achieving high-quality tumor segmentation. For the MRI dataset, the accurate segmentation rate of the model is 91.68%, which provides a certain reference for the medical field. In summary, this study demonstrates the usability of combining ResUNet50 and ResNet in brain tumor detection. In the field of medical image analysis, accurate brain tumor segmentation and classification are crucial for clinical diagnosis and treatment planning, and provide new ideas and methods for future medical image analysis and clinical applications.

Multi-branch spatial pyramid dynamic graph convolutional neural networks for solar defect detection

Multi-branch spatial pyramid dynamic graph convolutional neural networks for solar defect detection

Markerless vision-based knee osteoarthritis classification using machine learning and gait videos

IntroductionKnee osteoarthritis (KOA) is a major health issue affecting millions worldwide. This study employs machine learning algorithms to analyze human gait using kinematic data, aiming to enhance the diagnosis and detection of KOA. By adopting this approach, we contribute to the development of an effective diagnostic methods for KOA, a prevalent joint condition.MethodsThe methodology is structured around several critical steps to optimize the model’s performance. These steps include extracting kinematic features from video data to capture essential gait dynamics, applying data filtering and reduction techniques to remove noise and enhance data quality, and calculating key gait parameters to boost the model’s predictive power. The machine learning model trains on these refined features, validates through cross-validation for robust performance assessment, and tests on unseen data to ensure generalizability.ResultsOur approach demonstrates significant improvements in classification accuracy, highlighting its potential for early and precise KOA detection. The model achieves a high classification accuracy, indicating its effectiveness in distinguishing KOA-related gait patterns.DiscussionFurthermore, a comparative analysis with another model trained on the same dataset demonstrates the superiority of our method, suggesting that the proposed approach serves as a reliable tool for early KOA detection and potentially improves clinical diagnostic workflows.

Classification of hand movements from EEG using a FusionNet based LSTM network

Objective. Accurate classification of electroencephalogram (EEG) signals is crucial for advancing brain-computer interface (BCI) technology. However, current methods face significant challenges in classifying hand movement EEG signals, including effective spatial feature extraction, capturing temporal dependencies, and representing underlying signal dynamics.Approach. This paper introduces a novel multi-model fusion approach, FusionNet-Long Short-Term Memory (LSTM), designed to address these issues. Specifically, it integrates Convolutional Neural Networks for spatial feature extraction, Gated Recurrent Units and LSTM networks for capturing temporal dependencies, and Autoregressive (AR) models for representing signal dynamics.Main results. Compared to single models and state-of-the-art methods, this fusion approach demonstrates substantial improvements in classification accuracy. Experimental results show that the proposed model achieves an accuracy of 87.1% in cross-subject data classification and 99.1% in within-subject data classification. Additionally, Gradient Boosting Trees were employed to evaluate the significance of various EEG features to the model.Significance. This study highlights the advantages of integrating multiple models and introduces a superior classification model, which is pivotal for the advancement of BCI systems.

Radar Moving Target Detection Based on Small-Sample Transfer Learning and Attention Mechanism

Moving target detection is one of the most important tasks of radar systems. The clutter echo received by radar is usually strong and heterogeneous when the radar works in a complex terrain environment, resulting in performance degradation in moving target detection. Utilizing prior knowledge of the clutter distribution in the space–time domain, this paper proposes a novel moving target detection network based on small-sample transfer learning and attention mechanism. The proposed network first utilizes offline data to train the feature extraction network and reduce the online training time. Meanwhile, the attention mechanism used for feature extraction is applied in the beam-Doppler domain to improve classification accuracy of targets. Then, a small amount of real-time data are applied to a small-sample transfer network to fine-tune the feature extraction network. Finally, the target detection can be realized by the fine-tuned network. Simulation experiments show that the proposed network can eliminate the influence of heterogeneous clutter on moving target detection, and the attention mechanism can improve clutter suppression under a low signal-to-noise ratio regime. The proposed network has a lower computational load compared to conventional neural networks, enabling its use in real-time applications on space-borne/airborne radars.

Accurate Coal Classification Using PAIPSO-ELM with Near-Infrared Reflectance Spectroscopy.

China has vast proven coal reserves, encompassing a wide variety of types. However, traditional coal classification methods have limitations, often leading to inaccurate classification and inefficient utilization of coal resources. To address this issue, this paper introduces the Extreme Learning Machine (ELM) as a novel coal classification method, based on the near-infrared reflectance spectroscopy (NIRS) of coal. Initially, we collected NIRS data from coal samples using the SVC-HR-1024 spectrometer. Given the high dimensionality and strong linear correlations in NIRS data, we conducted preprocessing to enhance the usefulness of the data. In experiments, the ELM model demonstrated good classification performance. However, due to the random generation of input layer weights and hidden layer biases in the ELM model, its performance can be unstable, preventing the model from fully realizing its potential. To overcome this shortcoming, we employed the Particle Swarm Optimization (PSO) algorithm to optimize the parameters of the ELM model. Simulation results showed that the PSO-ELM model achieved a 9.68% improvement in classification accuracy compared to the original ELM model. Furthermore, we optimized the PSO algorithm by introducing exponentially decaying inertia factors and position-variant particles to further reduce the risk of the algorithm falling into local optima. The improved Position-Adaptive Inertia PSO-ELM (PAIPSO-ELM) model achieved an additional 2% increase in classification accuracy over the PSO-ELM model, without a significant increase in training time. In summary, this paper proposes a coal spectral classification method based on the PAIPSO-ELM model, effectively overcoming the limitations of traditional classification methods while meeting industrial demands for classification accuracy and speed.

MSCPUnet: A multi-task neural network for plot-level crop classification in complex agricultural areas

MSCPUnet: A multi-task neural network for plot-level crop classification in complex agricultural areas

Fine-tuning pre-trained networks with emphasis on image segmentation: A multi-network approach for enhanced breast cancer detection

Fine-tuning pre-trained networks with emphasis on image segmentation: A multi-network approach for enhanced breast cancer detection

Active Learning with Particle Swarm Optimization for Enhanced Skin Cancer Classification Utilizing Deep CNN Models.

Skin cancer is a critical global health issue, with millions of non-melanoma and melanoma cases diagnosed annually. Early detection is essential to improving patient outcomes, yet traditional deep learning models for skin cancer classification are often limited by the need for large, annotated datasets and extensive computational resources. The aim of this study is to address these limitations by proposing an efficient skin cancer classification framework that integrates active learning (AL) with particle swarm optimization (PSO). The AL framework selectively identifies the most informative unlabeled instances for expert annotation, minimizing labeling costs while optimizing classifier performance. PSO, a nature-inspired metaheuristic algorithm, enhances the selection process within AL, ensuring the most relevant data points are chosen. This method was applied to train multiple Convolutional Neural Network (CNN) models on the HAM10000 skin lesion dataset. Experimental results demonstrate that the proposed AL-PSO approach significantly improves classification accuracy, with the Least Confidence strategy achieving approximately 89.4% accuracy while using only 40% of the labeled training data. This represents a substantial improvement over traditional approaches in terms of both accuracy and efficiency. The findings indicate that the integration of AL and PSO can accelerate the adoption of AI in clinical settings for skin cancer detection. The code for this study is publicly available at ( https://github.com/Sayantani-31/AL-PSO ).

Capturing cognitive capacity in the everyday environment across a continuum of cognitive decline using a smartwatch n-back task and ecological momentary assessment.

Cognitive testing using mobile technologies can assist with early detection of cognitive decline. We use ecological momentary assessment to investigate the feasibility, reliability, and validity of a smartwatch n-back task (1-back) delivered to adults across the cognitive continuum. One hundred seventy-four community-dwelling individuals (Mage = 70.51) representing healthy older adults, individuals with subjective cognitive decline, and mild cognitive impairment completed a neurocognitive assessment battery and wore a smartwatch for 7+ days. Participants were prompted 4 times per day to complete an n-back task on the smartwatch. Across all groups, findings indicated an acceptable task adherence rate (> 78%; n = 174) and response rate (> 89%; n = 158 n-back analysis sample). Supporting external validity, participants with mild cognitive impairment were less accurate, had fewer total correct responses, and performed at lower initial levels than both healthy older adults and subjective cognitive decline, ω²s > .09. Intraindividual variability was greater for the mild cognitive impairment group compared to healthy older adults, but subjective cognitive decline did not differ significantly from either group, ω² = .12. For discriminant and convergent validity, n-back total correct correlated with performance on standardized assessments of executive attention, whereas intraindividual variability correlated with real-world factors (i.e., context, everyday function). Reliability assessment revealed stability for n-back measures after four to six posttraining trials and excellent test-retest reliability for total correct after 5 months. Finally, combining n-back and clinical measures improved classification accuracy. Findings suggest the smartwatch n-back task is feasible for collecting cognitive data across the cognitive continuum with demonstrated reliability and validity in the everyday environment using ecological momentary assessment. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

Abstract B010: Improving pancreatic cancer image classification with transfer learning: CD45 vs. PANCK

Abstract Pancreatic ductal adenocarcinoma (PDAC) is known for its high aggressiveness and is the fourth most common cause of death of cancer-related death in the United States. It is highly malignant and often goes undiagnosed until advanced stages due to its late-onset symptoms, resulting in poor prognosis. Using quantitative analyses of histological features, such as the expression of prognostic biomarker, can provide valuable insight to guide treatment decisions, monitor disease progression and predict outcomes. Previous researchers have found that CD45 and PANCK play a huge role in characterizing the tumor microenvironment of PDAC. One study using single-cell RNA sequencing compared the microenvironments of healthy pancreata and pancreatic tumors, revealing differences in cellular composition. Tumor samples showed a higher proportion of immune cells, including CD45+ cells, compared to heathy samples and utilized PANCK to differentiate between epithelial and non-epithelial cells. This study aims to enhance the accuracy of classifying pancreatic cancer pathological images based on prognostic markers using transfer learning techniques. We curated and pre-processed a dataset of pancreatic cancer pathological images with different prognostic markers. Using the pre-trained convolutional neural network (CNN) model VGG16, we fine-tuned it with our dataset to better differentiate between two prognostic markers, CD45 and PANCK. The model's performance was evaluated using a similarity matrix to illustrate distinctions among the two prognostic markers. Our results demonstrate that the transfer learning approach significantly improves classification accuracy, achieving confusion matrix validation results of 90% between CD45 and PANCK prognostic markers. SSIM values were relatively low, highlighting the ability of our model to differentiate between the two prognostic markers. The low SSIM value also shows variability within the two prognostic markers, indicating heterogeneity in structural changes due to pancreatic cancer. These findings indicate that transfer learning can be a powerful tool for improving the classification of pathological images and that there is notable variability in the structural features associated with different prognostic markers in pancreatic cancer. Citation Format: Miracle Thomas. Improving pancreatic cancer image classification with transfer learning: CD45 vs. PANCK [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Tumor-body Interactions: The Roles of Micro- and Macroenvironment in Cancer; 2024 Nov 17-20; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2024;84(22_Suppl):Abstract nr B010.