Yazar "Tasci, Erdal" seçeneğine göre listele
Listeleniyor 1 - 12 / 12
Sayfa Başına Sonuç
Sıralama seçenekleri
Öğe Bias and Class Imbalance in Oncologic Data-Towards Inclusive and Transferrable AI in Large Scale Oncology Data Sets(Mdpi, 2022) Tasci, Erdal; Zhuge, Ying; Camphausen, Kevin; Krauze, Andra V.Simple Summary Large-scale medical data carries significant areas of underrepresentation and bias at all levels: clinical, biological, and management. Resulting data sets and outcome measures reflect these shortcomings in clinical, imaging, and omics data with class imbalance emerging as the single most significant issue inhibiting meaningful and reproducible conclusions while impacting the transfer of findings between the lab and clinic and limiting improvement in patient outcomes. When employing artificial intelligence methods, class imbalance can produce classifiers whose predicted class probabilities are geared toward the majority class ignoring the significance of minority classes, in turn generating algorithmic bias. The inability to mitigate this can guide an AI system in favor of or against various cohorts or variables. We review sources of bias and class imbalance and relate this to AI methods. We discuss avenues to mitigate these and propose a set of guidelines aimed at limiting and addressing data and algorithmic bias. Recent technological developments have led to an increase in the size and types of data in the medical field derived from multiple platforms such as proteomic, genomic, imaging, and clinical data. Many machine learning models have been developed to support precision/personalized medicine initiatives such as computer-aided detection, diagnosis, prognosis, and treatment planning by using large-scale medical data. Bias and class imbalance represent two of the most pressing challenges for machine learning-based problems, particularly in medical (e.g., oncologic) data sets, due to the limitations in patient numbers, cost, privacy, and security of data sharing, and the complexity of generated data. Depending on the data set and the research question, the methods applied to address class imbalance problems can provide more effective, successful, and meaningful results. This review discusses the essential strategies for addressing and mitigating the class imbalance problems for different medical data types in the oncologic domain.Öğe Development of a Novel Feature Weighting Method Using CMA-ES Optimization(Ieee, 2018) Tasci, Erdal; Gokalp, Osman; Ugur, AybarsFeature weighting is one of the fundamental problems in machine learning algorithms and data mining to determine the importance of features. in this study, a novel feature weighting method using Covariance Matrix Adaptation Evolution Strategy (CMA-ES) optimization method for classification process is proposed. Experimental results are obtained by 10-fold cross validation technique with 3 different classifier models: Naive Bayes (NB), K nearest neighbors (K-NN) and Random Forest (RF) on 5 datasets in UCI Machine Learning Repository. Classification accuracy rate is used as the performance criterion. in addition, the developed CMAES-based method is also adapted to optimize these 3 classifiers in a voting-based ensemble algorithm. in this context, a different ensemble-based method is presented with CMAES-based feature weights obtained when the classifiers are individually and all together. Experimental studies show that the developed method gives better performance and promising results than the results obtained without feature weighting.Öğe Image Classification Using Ensemble Algorithms with Deep Learning and Hand-Crafted Features(Ieee, 2018) Tasci, Erdal; Ugur, AybarsIn this study, ensemble learning based image classification method is proposed by using both features extracted by means of pre-trained convolutional neural networks (CNN) and hand-crafted. Recently, deep learning models have been widely used in computer vision applications and significantly increase performance. in this scope, classification process is performed by adding 4 hand-crafted features to 4096 deep learning features on the CIFAR-10 dataset. The contribution to the performance of system is measured by using both hand-crafted and deep learning features together. Classification accuracy rate is used as the performance criterion. Experimental studies show that the developed method gives better results than only using the deep learning features.Öğe A novel pattern recognition framework based on ensemble of handcrafted features on images(Springer, 2022) Tasci, Erdal; Ugur, AybarsNowadays, with the advances and use of technological possibilities and devices, the number of digital images is increasing gradually. Computer-aided classification of image types is widely applied in many applications such as medicine, security, and automation. The feature extraction and selection stages have great importance in terms of improving the classification performance as sub-stages of the pattern recognition process. Researchers apply different feature extraction methods for their works due to the requirements. In this study, a novel pattern recognition framework combining diverse and large-scale handcrafted feature extraction methods (shape-based and texture-based) and the selection stage on images is developed. Genetic algorithms are also used for feature selection. In the experimental studies, Flavia leaf recognition, Caltech101 object classification image datasets, and five supervised classification models (random forest, ECOC-SVM, k-nearest neighbor, AdaBoost, classification tree) with different parameters' values are used. The experimental results show that the proposed method achieves 98.39% and 82.77% accuracy rates on Flavia and Caltech101 datasets with the ECOC-SVM model, respectively. The proposed framework is also competitive with the existing state-of-the-art methods in the related literature.Öğe A novel wrapper feature selection algorithm based on iterated greedy metaheuristic for sentiment classification(Pergamon-Elsevier Science Ltd, 2020) Gokalp, Osman; Tasci, Erdal; Ugur, AybarsIn recent years, sentiment analysis is becoming more and more important as the number of digital text resources increases in parallel with the development of information technology. Feature selection is a crucial sub-stage for the sentiment analysis as it can improve the overall predictive performance of a classifier while reducing the dimensionality of a problem. in this study, we propose a novel wrapper feature selection algorithm based on Iterated Greedy (IG) metaheuristic for sentiment classification. We also develop a selection procedure that is based on pre-calculated filter scores for the greedy construction part of the IG algorithm. A comprehensive experimental study is conducted on commonly-used sentiment analysis datasets to assess the performance of the proposed method. the computational results show that the proposed algorithm achieves 96.45% and 90.74% accuracy rates on average by using Multi-nomial Naive Bayes classifier for 9 public sentiment and 4 Amazon product reviews datasets, respectively. the results also reveal that our algorithm outperforms state-of-the-art results for the 9 public sentiment datasets. Moreover, the proposed algorithm produces highly competitive results with state-of-the-art feature selection algorithms for 4 Amazon datasets. (C) 2020 Elsevier Ltd. All rights reserved.Öğe Pre-processing Effects of the Tuberculosis Chest X-Ray Images on Pre-trained CNNs: An Investigation(Springer International Publishing Ag, 2020) Tasci, ErdalTuberculosis (TB) is a serious infectious disease which is one of the top causes of death worldwide. In 2017, 1.6 million people died from the disease according to the World Health Organization (WHO). The earlier identification and treatment of the TB is critical for preventing death and decreasing risk of transmitting the disease to others. Computer-aided diagnosis (CADx) systems are essential tools to speed up the decision-making process of experts and provide more efficient, accurate and systematic solutions. Chest radiography (CXR) is one of the most common and effective imaging technique for the detection of thoracic diseases such as TB and lung cancer. In this study, three different region of interests (ROIs) based pre-processing methods are applied to two CXR image datasets (namely, Montgomery and Shenzhen). We used three pre-trained convolutional neural networks (CNNs) (namely, AlexNet, VGG16, VGG19) as deep learning models and deep feature extractors for automatic classification of TB disease. We investigate the pre-processing effects of TB CXR images on the classifier whether ROI is selected and remaining regions of images are set pixel values to white, black and same pixel values in the original images. Experimental results indicate that proposed methods contribute to the classifier performance gain considerably in terms of accuracy rate.Öğe Shape and Texture Based Novel Features for Automated Juxtapleural Nodule Detection in Lung CTs(Springer, 2015) Tasci, Erdal; Ugur, AybarsLung cancer is one of the types of cancer with highest mortality rate in the world. In case of early detection and diagnosis, the survival rate of patients significantly increases. In this study, a novel method and system that provides automatic detection of juxtapleural nodule pattern have been developed from cross-sectional images of lung CT (Computerized Tomography). Shape-based and both shape and texture based 7 features are contributed to the literature for lung nodules. System that we developed consists of six main stages called preprocessing, lung segmentation, detection of nodule candidate regions, feature extraction, feature selection (with five feature ranking criteria) and classification. LIDC dataset containing cross-sectional images of lung CT has been utilized, 1410 nodule candidate regions and 40 features have been extracted from 138 cross-sectional images for 24 patients. Experimental results for 10 classifiers are obtained and presented. Adding our derived features to known 33 features has increased nodule recognition performance from 0.9639 to 0.9679 AUC value on generalized linear model regression (GLMR) for 22 selected features and being reached one of the most successful results in the literature.Öğe Shape and Texture Based Novel Features for Automated Juxtapleural Nodule Detection in Lung CTs(Springer, 2015) Tasci, Erdal; Ugur, AybarsLung cancer is one of the types of cancer with highest mortality rate in the world. In case of early detection and diagnosis, the survival rate of patients significantly increases. In this study, a novel method and system that provides automatic detection of juxtapleural nodule pattern have been developed from cross-sectional images of lung CT (Computerized Tomography). Shape-based and both shape and texture based 7 features are contributed to the literature for lung nodules. System that we developed consists of six main stages called preprocessing, lung segmentation, detection of nodule candidate regions, feature extraction, feature selection (with five feature ranking criteria) and classification. LIDC dataset containing cross-sectional images of lung CT has been utilized, 1410 nodule candidate regions and 40 features have been extracted from 138 cross-sectional images for 24 patients. Experimental results for 10 classifiers are obtained and presented. Adding our derived features to known 33 features has increased nodule recognition performance from 0.9639 to 0.9679 AUC value on generalized linear model regression (GLMR) for 22 selected features and being reached one of the most successful results in the literature.Öğe Shape and Texture Based Novel Features for Automated Juxtapleural Nodule Detection in Lung CTs(Springer, 2015) Tasci, Erdal; Ugur, AybarsLung cancer is one of the types of cancer with highest mortality rate in the world. In case of early detection and diagnosis, the survival rate of patients significantly increases. In this study, a novel method and system that provides automatic detection of juxtapleural nodule pattern have been developed from cross-sectional images of lung CT (Computerized Tomography). Shape-based and both shape and texture based 7 features are contributed to the literature for lung nodules. System that we developed consists of six main stages called preprocessing, lung segmentation, detection of nodule candidate regions, feature extraction, feature selection (with five feature ranking criteria) and classification. LIDC dataset containing cross-sectional images of lung CT has been utilized, 1410 nodule candidate regions and 40 features have been extracted from 138 cross-sectional images for 24 patients. Experimental results for 10 classifiers are obtained and presented. Adding our derived features to known 33 features has increased nodule recognition performance from 0.9639 to 0.9679 AUC value on generalized linear model regression (GLMR) for 22 selected features and being reached one of the most successful results in the literature.Öğe Voting combinations-based ensemble of fine-tuned convolutional neural networks for food image recognition(Springer, 2020) Tasci, ErdalObesity is one of today's most visible, uncared, and common public health problems worldwide. To manage weight loss, obtain calorie intake and record eating lists, the development of the diverse automatic dietary assessment applications has great importance. Recently, deep learning becomes a popular approach that provides outstanding image recognition results. in this paper, we use ResNet, GoogleNet, VGGNet, and InceptionV3 with fine-tuning based on deep learning for image-based and computer-aided food recognition task. We also apply six voting combination rules (namely, minimum probability, average of probabilities, median, maximum probability, product of probabilities, and weighted probabilities) for ensemble methods. the experimental results demonstrate that our proposed ensemble voting scheme with transfer learning gives promising results compared to the state-of-the-art methods on Food-101, UEC-FOOD100, and UEC-FOOD256 image datasets.Öğe A voting-based ensemble deep learning method focusing on image augmentation and preprocessing variations for tuberculosis detection(Springer London Ltd, 2021) Tasci, Erdal; Uluturk, Caner; Ugur, AybarsTuberculosis (TB) is known as a potentially dangerous and infectious disease that affects mostly lungs worldwide. The detection and treatment of TB at an early stage are critical for preventing the disease and decreasing the risk of mortality and transmission of it to others. Nowadays, as the most common medical imaging technique, chest radiography (CXR) is useful for determining thoracic diseases. Computer-aided detection (CADe) systems are also crucial mechanisms to provide more reliable, efficient, and systematic approaches with accelerating the decision-making process of clinicians. In this study, we propose voting and preprocessing variations-based ensemble CNN model for TB detection. We utilize 40 different variations in fine-tuned CNN models based on InceptionV3 and Xception by also using CLAHE (contrast-limited adaptive histogram equalization) preprocessing technique and 10 different image transformations for data augmentation types. After analyzing all these combination schemes, three or five best classifier models are selected as base learners for voting operations. We apply the Bayesian optimization-based weighted voting and the average of probabilities as a combination rule in soft voting methods on two TB CXR image datasets to get better results in various numbers of models. The computational results indicate that the proposed method achieves 97.500% and 97.699% accuracy rates on Montgomery and Shenzhen datasets, respectively. Furthermore, our method outperforms state-of-the-art results for the two TB detection datasets in terms of accuracy rate.Öğe Weighted Voting Based Ensemble Classification with Hyper-parameter Optimization(Ieee, 2019) Gokalp, Osman; Tasci, ErdalEnsemble learning is one of the most popular research fields in machine learning and pattern recognition due to its contribution to the performance of a classification system. Voting based ensemble methods employs multiple learning algorithms and make the classification model more robust. Weighted voting based ensemble methods provide more flexible and fine-grained way to predict actual output classes compared to the unweighted (majority) voting based ensemble methods. However, the main drawback of weighted methods is to decide which values to be used. in this study, two hyper-parameter optimization strategies, namely Random Search and TPE, are used for optimizing weights of voting based ensemble classification. Experimental results based on seven machine learning datasets demonstrate the effectiveness of using hyper-parameter optimization for the purpose of finding optimal values for weighted voting based ensemble classification.
