Yazar "Gundogmus, Yunus Emre" seçeneğine göre listele

Listeleniyor 1 - 2 / 2
  • Küçük Resim Yok
    Öğe
    A Glimpse to Turkish Political Climate with Statistical Machine Learning
    (IEEE, 2018) Cetin, Uzay; Gundogmus, Yunus Emre
    In this study, we conduct a data-driven study to harvest the decision makers' policy orientation and predict their votes. We collect and analyze the data about the opinion of the individual voters on a variety of political issues related to Turkish politics. Based on this data, we can measure which parties are close and which parties are distant in multi-dimensional political space. We can make a glimpse to what social matters shape the Turkish political climate with the lenses of statistical models. We show in which political issues Turkish people agree on the most and in which political issues they are segregated the most. Moreover, by using traditional machine learning tools, we try to predict the vote of an individual, depending on his or her opinion about the pre-determined political issues with the help of our data.
  • Küçük Resim Yok
    Öğe
    Feature Selection with Evolving, Fast and Slow Using Two Parallel Genetic Algorithms
    (IEEE, 2019) Cetin, Uzay; Gundogmus, Yunus Emre
    Feature selection is one of the most challenging issues in machine learning, especially while working with high dimensional data. In this paper, we address the problem of feature selection and propose a new approach called Evolving Fast and Slow. This new approach is based on using two parallel genetic algorithms having high and low mutation rates, respectively. Evolving Fast and Slow requires a new parallel architecture combining an automatic system that evolves fast and an effortful system that evolves slow. With this architecture, exploration and exploitation can be done simultaneously and in unison. Evolving last, with high mutation rate, can be useful to explore new unknown places in the search space with long jumps; and Evolving Slow, with low mutation rate, can be useful to exploit previously known places in the search space with short movements. Our experiments show that Evolving Fast and Slow achieves very good results in terms of both accuracy and feature elimination.