Uzaktan Eğitim Uygulama ve Araştırma Merkezi Yayın Koleksiyonu
Bu koleksiyon için kalıcı URI
Güncel Gönderiler
Öğe Baby Face Generation with Generative Adversarial Neural Networks: A Case Study(2020) Ortaç, Gizem; Doğan, Zeliha; Orman, Zeynep; Şamlı, RüyaGenerative Adversarial Networks (GANs) are increasingly applied to train generative models with neural networks, especially in computer vision studies. Since being introduced in 2014, many image generation studies incorporating GANs have demonstrated promising results for producing highly convincing fake images of animals, landscapes, and human faces. We build a GAN structure to generate realistic baby face images from a small data set of 673 color 200×200 pixel images obtained from a Kaggle data set by following previous studies that demonstrated how GANs could be used for image generation from a limited number of training samples. The reason we limit especially as baby faces is that we aim to achieve success with a limited number of training data. For evaluation, experiments and case studies are one of the most considered techniques. The results of this study help identify issues requiring further investigation in comment analysis research. In this context, we presented the loss values of the generator and discriminator during the training process. The discriminator losses are around of 0.7 and the generator is between 0.7 and 0.9. The high quality images are produced about 300th epochs.Öğe A COMPARATIVE STUDY FOR HYPERSPECTRAL DATA CLASSIFICATION WITH DEEP LEARNING AND DIMENSIONALITY REDUCTION TECHNIQUES(2018) Ortaç, Gizem; Özcan, GıyasettinIn recent years, hyperspectral imaging has been a popular subject in the remote sensing community by providing a rich amount of information for each pixel about fields. In general, dimensionality reduction techniques are utilized before classification in statistical pattern-classification to handle high-dimensional and highly correlated feature spaces. However, traditional classifiers and dimensionality reduction methods are difficult tasks in the spectral domain and cannot extract discriminative features. Recently, deep convolutional neural networks are proposed to classify hyperspectral images directly in the spectral domain. In this paper, we present comparative study among traditional data reduction techniques and convolutional neural network. The obtained results on hyperspectral image data sets show that our proposed CNN architecture improves the accuracy rates for classification performance, when compared to traditional methods by increasing the classification accuracy rate by 3% and 6%.
