Görüntüleme
12
İndirme
4
Parkinson Hastalığı (PH), bireylerin çoklu motor ve motor olmayan özelliklerini doğrudan etkileyen ilerleyici bir sinir hastalığıdır. PH’nin ilk evresinde bireyler genellikle ses bozulmalarıyla karşı karşıya kalır. Bu durumda PH’nin erken tespitinde kişilerin ses kayıtlarından yararlanılır. Ses kayıtlarından sinyal işleme yöntemleriyle çıkarılan öznitelikler yapay öğrenme yöntemlerine girdi olarak verilerek bireylerin hastalığa sahip olup olmadığı tespit edilir. Bu çalışmada bireylerin ses kayıtlarından çıkarılan öznitelikler iki farklı yapay öğrenme yöntemine girdi olarak verilmiş ve bireyler Parkinson hastası veya sağlıklı olarak sınıflandırılmıştır. Oluşturulan modeller UCI Makine Öğrenmesi deposundan alınan veri kümesi ile eğitilmiştir. Hem eğitilen yapay öğrenme modellerinin karmaşıklığını azaltmak hem de modellerin aşırı öğrenmesini engellemek için öznitelikler üzerinde iki farklı boyutsallık indirgeme yöntemi uygulanmıştır. İlk yöntem olan Temel Bileşenler Analizi (TBA)’yle yeni bir öznitelik alt uzayı oluşturmak için öznitelik kümesi orijinal boyuttan daha az boyuta sahip olan yeni bir alt uzaya yansıtılır. Oluşturulan yeni öznitelik uzayında yüksek varyansa sahip bileşenler seçilirken varyansı düşük bileşenler ihmal edilir. İkinci yöntem olan Özyinelemeli Öznitelik Eleme (ÖÖE)’de özniteliklere yapay öğrenme yöntemleri kullanılarak ilgililik puanları atanır. İlk aşamada tüm öznitelik kümesini kullanan bir model oluşturulur ve her öznitelik için bir ilgililik puanı hesaplanır. Sonraki aşamada en az ilgililik puanına sahip öznitelik ihmal edilerek model yeniden oluşturulur ve ilgililik puanları tekrar hesaplanır. Bu işlem öznitelik kümesinde istenilen sayıda öznitelik kalana kadar devam ettirilir. Kullanlan iki Boyutsallık indirgeme yöntemiyle öznitelik uzayının boyutları azaltılmış ve indirgenmiş öznitelik vektörleriyle Destek Vektör Makineleri (DVM) ve Gradyan Arttırıcı Makineler (GAM) sınıflandırıcıları eğitilmiştir. Elde edilen veri kümesinin örnek sayısı görece az olduğundan sınıflandırıcıların eğitiminde Bireyi Dışarda Bırakan Çapraz Doğrulama (BDBÇD) prosedürü kullanılmıştır. Veri kümesi aynı zamanda dengesiz sınıf dağılımına sahip olduğundan modellerin performans değerlendirmesinde doğruluk oranıyla birlikte F-ölçütü ve Matthews Korelasyon Katsayısı (MKK) ölçütleri kullanılmıştır. Alınan tüm deneysel sonuçlar irdelendiğinde en yüksek sınıflandırma başarısına sadece 13 öznitelik kullanılarak erişildiği görülmüştür. ÖÖE yöntemiyle seçilen 13 öznitelikle GAM sınıflandırıcısı eğitilerek 0,881 doğruluk oranı elde edilmiştir. Doğruluk oranı öznitelik seçimi yapılmadan elde edilen sonuçlara göre yaklaşık %2 oranında artmıştır. Aynı artış sınıfların ayırt edilebilirliğini gösteren MKK oranında da olmuştur. Boyutsallık indirgeme işlemi olmadan elde edilen MKK oranı 0,62 iken ÖÖE yöntemiyle öznitelik seçimi yapıldığında oran 0,67’ye yükselmiştir. Kullanılan diğer boyutsallık indirgeme yöntemi olan TBA ise öznitelik seçimsiz modellere göre sınıflandırma başarısı arttırmamasına rağmen, aynı başarı oranlarına daha az sayıda öznitelikle erişmiştir.
Parkinson's disease (PH) is a progressive nerve disease that directly affects the multi-motor and non-motor properties of individuals. In the first phase of PH, individuals often face noise disorders. In this case, the early detection of PH is used by people's audio recordings. The properties extracted from sound recordings by means of signal processing methods are given as an input to artificial learning methods to identify whether individuals have a disease. In this study, the properties extracted from the audio records of individuals were given as input to two different methods of artificial learning and individuals were classified as Parkinson's or healthy. Created models are trained with the data set obtained from the UCI Machine Learning warehouse. Two different measurement reduction methods have been applied on properties to reduce the complexity of both trained artificial learning models and to prevent models from excessive learning. The first method is Basic Components Analysis (TBA) to create a new subspace of the subspace of the subspace of the subspace of the subspace of the subspace of the subspace of the subspace of the subspace of the subspace of the subspace of the subspace. The components with high variants are selected in the new characteristic space created; the components with low variance are ignored. The second method is the self-released self-released self-released self-released self-released self-released self-released self-released self-released self-released. In the first stage, a model is created using the whole set of properties and a relevance score for each properties is calculated. In the next step, the subjectivity with the minimum interest score is ignored and the model is re-created and the interest points are re-calculated. This process will continue until the desired number of properties remains. The two-dimensional reduction method used reduced and reduced dimensions of the specificity space with reduced specificity vectors support vector machines (DVMs) and gradyan boosting machines (GAMs) classifiers. Since the number of samples of the obtained data set is relatively small, the classifiers’ training has used the Interface Verification (BDBÇD) procedure that leaves the individual out. Since the data set also has an unbalanced class distribution, the performance assessment of the models has been used with the accuracy ratio of the F-dimensional and Matthews Correlation Scorecard (MKK) standards. When all the results were obtained, the highest classification success was found to be achieved using only 13 properties. With the 13 subjects selected by the GAM classifier, the 0.881 accuracy ratio was obtained. The accuracy rate increased by approximately 2 percent compared to the results obtained without the selection. The same increase has also occurred in the rate of MKK that shows the distinctivity of classes. The MKK ratio obtained without the measurement reduction process was 0.62, while the ratio increased to 0.67 when the specificity was selected by the OEE method. The other measurement reduction method used is the TBA, which has not increased the classification success in comparison with the non-selective models, but has reached the same success rates with a fewer number of classifications.
Parkinson's Disease (PD) is a progressive neural disease that directly affects multiple motor and non-motor features of the individuals. PD individuals are often confronted with sound distortion in the first stage of the disease. In this case, the voice recordings of the people are used for the early detection. The features extracted from the sound recordings by signal processing methods are given as input to machine learning methods for the detection of the PD. In this study, the features extracted from the voice recordings of individuals were given as input to two different machine learning models for the detection of PD. The models were trained with the dataset obtained from the UCI Machine Learning repository. Two different dimensionality reduction methods were applied on the features in order to reduce the complexity of the trained models and to prevent the over-fitting. The first method, Principal Components Analysis (TBA), projects original feature space into a new subspace that has fewer dimensions than the original. In order to reduce feature dimensions, components with high variances in the new feature space are selected. In the second method, Recursive Feature Elimination (RFE), relevance scores are assigned to the features by using machine learning methods. In the first step, a model that uses the entire set of features is created and a relevance score is calculated for each feature. In the next stage, the model is rebuilt by neglecting the feature with the least relevance score and the relevance scores are recalculated. This process is continued until the desired number of features remains in the feature set. After dimensionality reduction process, Support Vector Machines (SVM) and Gradient Boosting Machines (GBM) classifiers were trained with selected features. Since the number of intances in the dataset was small, One Person Out Cross Validation (OPOCV) was used in classifier training. Due to having data imbalance problem, F-Measure and Matthews Correlation Coefficient (MCC) metrics were used along with accuracy in the performance evaluation. When all the experimental results were examined, it was found out that the highest classification success was achieved by using only 13 features. The GBM classifier was trained with 13 features selected by RFE method to obtain an accuracy of 0.881. Accuracy rate increased by about 2% according to the results obtained without feature selection. The same increase was also seen in the rate of MCC that shows the degree of the class distinguishability. While MCC rate obtained without dimensionality reduction was 0.62, the ratio increased to 0.67 when the feature selection was done with the RFE. PCA, which is the other dimensionality reduction method, did not increase the classification success compared to without selection, but achieved the same success rates with fewer features.
Alan : Fen Bilimleri ve Matematik; Mühendislik
Dergi Türü : Uluslararası
Makale Detay 
Benzer Makaleler
PDF Görüntüle
Dergi Bilgisi
Eseri Dinleyin
Alıntı Yap
Bu Sayfayı Yazdırın
Paylaş
Mendeley
