Qualifications required for prospective graduate students

My research agenda focuses on high-dimensional data analysis problems where the number of features are large in number compared to the number of data points. I am interested in integrating convex or non-convex penalty functions such as SCAD into some supervised learning algorithms (regression & classification problems). If you are interested in pursuing a graduate study with me, you should already be familiar with generalized linear models (linear regression, logistic regression), lasso regression, ridge regression, and have very good programming skills. I can briefly list the requirements you need to have prior to asking for an appointment from me:

  • Previous course work on: Linear algebra, optimization, probability theory, statistics (multiple statistics courses), and machine learning.
  • Good level of Python knowledge.
  • Familiarity with Git/GitHub and LaTeX (and/or Overleaf).
  • Being a research-oriented person, being able to read and critize research papers published in machine learning venues from a mathematical/statistical/computational view point.
  • Strong communication skills (both verbally and orally).

Thank you very much for your interest. I am planning to hire 1-2 grad students to work on developing statistical methodology for high-dimensional data analysis in Spring23. Unfortunately, I am not looking for undergrad students until Fall23.


Supervision

  • Master Thesis Project:
    • İrem Sarıbaş. High-dimensional data analysis.
    • Ozan Güldali (August 2021). Deep feature transfer from deep learning models into machine learning algorithms to classify Covid-19 from chest X-ray images. [Thesis], [Code].
  • Undergraduate Senior Thesis Project:
    • Yusuf Emre Bayat, Tarık Kayalı, Umut Özbağrıaçık "Anomaly detection in industrial data" (Start: July 2022). We acknowledge the support of ITU Technopolis firm Sensemore operating on machine health monitoring and failure analysis.
    • İpek Korkmaz, Eda Atalay, Burcu Cebecioğlu, Şeval Çoşar, "Recurrent neural networks with applications" (June 2022).
    • Sercan Yıldız, "Data science tools for mining digitized historical archives" (August 2021). Supported by Tubitak 2209-A.
    • Paşa Çağrı Köroğlu, "Investigating Early Republican Era Mathematicians of Turkey, namely, Cahit Arf, Mustafa İnan, Kerim Erim, Ratip Berker, and Von Mises in Digitized Historical Archives" (August 2021). Supported by Tubitak 2209-A.
    • Nemat J. Jahangirov (January 2021).
    • Sami Aldağ, Doğukan Topçuoğlu, Bertan Taylan, Hakan Turgay (January 2020).
    • Erol Kıbrıs (June 2019).