Bilgisayar Mühendisliği Bölümü Koleksiyonu

Permanent URI for this collectionhttps://hdl.handle.net/20.500.11779/1940

Browse

Filters

20212

Settings

Subject: search.filters.subject.Transfer learning

Search Results

Now showing 1 - 2 of 2
  • Conference Object
    Citation - WoS: 1
    Citation - Scopus: 2
    Turcoins: Turkish Republic Coin Dataset
    (IEEE, 2021-06-09) Gökberk, Berk; Akarun, Lale; Temiz, Hüseyin
    In this paper, we present a novel and comprehensive dataset which contains Turkish Republic coins minted since 1924 and present a deep learning based system that can automatically classify coins. The proposed dataset consists of 11080 coin images from 138 different classes. To classify coins, we utilize a pre-trained neural network (ResNet50) which is pre-trained on ImageNet. We train the pre-trained neural networks on our dataset by transfer learning. The imbalanced nature of the dataset causes the classifier to show lower performance in classes with fewer samples. To alleviate the imbalance problem, we propose a StyleGAN2-based augmentation method providing realisticfake coins for rare classes. The dataset will be published in http://turcoins.
  • Article
    Citation - WoS: 19
    Citation - Scopus: 28
    An Evaluation of Recent Neural Sequence Tagging Models in Turkish Named Entity Recognition
    (Elsevier, 2021-11-01) Makaroğlu, Didem; Demir, Şeniz; Aras, Gizem; Çakır, Altan
    Named entity recognition (NER) is an extensively studied task that extracts and classifies named entities in a text. NER is crucial not only in downstream language processing applications such as relation extraction and question answering but also in large scale big data operations such as real-time analysis of online digital media content. Recent research efforts on Turkish, a less studied language with morphologically rich nature, have demonstrated the effectiveness of neural architectures on well-formed texts and yielded state-of-the art results by formulating the task as a sequence tagging problem. In this work, we empirically investigate the use of recent neural architectures (Bidirectional long short-term memory (BiLSTM) and Transformer-based networks) proposed for Turkish NER tagging in the same setting. Our results demonstrate that transformer-based networks which can model long-range context overcome the limitations of BiLSTM networks where different input features at the character, subword, and word levels are utilized. We also propose a transformer-based network with a conditional random field (CRF) layer that leads to the state-of-the-art result (95.95% f-measure) on a common dataset. Our study contributes to the literature that quantifies the impact of transfer learning on processing morphologically rich languages.