Tuğba Erkoç
Research Subjects
Tuğba Erkoç has a total of 5 co-authors. Here are the top co-authors by number of shared publications:
- Didem Ağdoğan
Pattern Recognition and Machine Intelligence Laboratory, IŞIK University, Şile, Istanbul, Turkey
1 shared publication K.K. Emre Aki
Pattern Recognition and Machine Intelligence Laboratory, Department of Computer Engineering, Işık University, İstanbul, Turkey
1 shared publicationM. Taner Eski
Pattern Recognition and Machine Intelligence Laboratory, Department of Computer Engineering, Işık University, İstanbul, Turkey
1 shared publication- M. Taner Eskil
Isik University, Istanbul, Turkey
1 shared publication Mustafa Taner Eskıl
Pattern Recognition and Machine Intelligence Laboratory, Ik University, ile, Istanbul, Turkey
1 shared publication
Latest Publications
- Open Access
- 17 May 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 49393-49408
A Novel Similarity Based Unsupervised Technique for Training Convolutional Filters
Achieving satisfactory results with Convolutional Neural Networks (CNNs) depends on how effectively the filters are trained. Conventionally, an appropriate number of filters is carefully selected, the filters are initialized with a proper initialization method and trained with backpropagation over several epochs. This training scheme requires a large labeled dataset, which is costly and time-consuming to obtain. In this study, we propose an unsupervised approach that extracts convolutional filters from a given dataset in a self-organized manner by processing the training set only once without using backpropagation training. The proposed method allows for the extraction of filters from a given dataset in the absence of labels. In contrast to previous studies, we no longer need to select the best number of filters and a suitable filter weight initialization scheme. Applying this method to the MNIST, EMNIST-Digits, Kuzushiji-MNIST, and Fashion-MNIST datasets yields high test performances of 99.19%, 99.39%, 95.03%, and 90.11%, respectively, without applying backpropagation training or using any preprocessed and augmented data.
- 17 May 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 49393-49408
Open AccessA Novel Similarity Based Unsupervised Technique for Training Convolutional Filters
Achieving satisfactory results with Convolutional Neural Networks (CNNs) depends on how effectively the filters are trained. Conventionally, an appropriate number of filters is carefully selected, the filters are initialized with a proper initialization method and trained with backpropagation over several epochs. This training scheme requires a large labeled dataset, which is costly and time-consuming to obtain. In this study, we propose an unsupervised approach that extracts convolutional filters from a given dataset in a self-organized manner by processing the training set only once without using backpropagation training. The proposed method allows for the extraction of filters from a given dataset in the absence of labels. In contrast to previous studies, we no longer need to select the best number of filters and a suitable filter weight initialization scheme. Applying this method to the MNIST, EMNIST-Digits, Kuzushiji-MNIST, and Fashion-MNIST datasets yields high test performances of 99.19%, 99.39%, 95.03%, and 90.11%, respectively, without applying backpropagation training or using any preprocessed and augmented data.
- 1 May 2018
- Published by Elsevier in Signal Processing: Image Communication
- Vol. 64, 11-20
An observation based muscle model for simulation of facial expressions
- 1 May 2018
- Published by Elsevier in Signal Processing: Image Communication
- Vol. 64, 11-20
An observation based muscle model for simulation of facial expressions
- 1 December 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 144-147
Subset selection for tuning of hyper-parameters in artificial neural networks
Hyper-parameters of a machine learning architecture define its design. Tuning of hyper-parameters is costly and for large data sets outright impractical, whether it is performed manually or algorithmically. In this study we propose a Neocognitron based method for reducing the training set to a fraction, while keeping the dynamics and complexity of the domain. Our approach does not require processing of the entire training set, making it feasible for larger data sets. In our experiments we could successfully reduce the MNIST training data set to less than 2.5% (1,489 images) by processing less than 10% of the 60K images. We showed that the reduced data set can be used for tuning of number of hidden neurons in a multi-layer perceptron.
- 1 December 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 144-147
Subset selection for tuning of hyper-parameters in artificial neural networks
Hyper-parameters of a machine learning architecture define its design. Tuning of hyper-parameters is costly and for large data sets outright impractical, whether it is performed manually or algorithmically. In this study we propose a Neocognitron based method for reducing the training set to a fraction, while keeping the dynamics and complexity of the domain. Our approach does not require processing of the entire training set, making it feasible for larger data sets. In our experiments we could successfully reduce the MNIST training data set to less than 2.5% (1,489 images) by processing less than 10% of the 60K images. We showed that the reduced data set can be used for tuning of number of hidden neurons in a multi-layer perceptron.