Grega Vrbančič is a Young Researcher, Ph.D. Student and a member of Institute of Informatics at the University of Maribor, Faculty of Electrical Engineering and Computer Science. His primary research interests include Artificial Intelligence and Machine Learning - especially area of Deep Learning.


  • Artificial Intelligence
  • Machine Learning
  • Deep Learning
  • Data Science


  • MEng in Information Technology, 2017

    University of Maribor, Faculty of Electrical Engineering and Computer Science

  • BSc in Information Technology, 2015

    University of Maribor, Faculty of Electrical Engineering and Computer Science

My writings

Keras with Tensorflow - How to flush CUDA Memory

As a young researcher working in the field of machine learning, I am on a regular basis utilizing the GPUs to train different kinds of neural networks such as convolutional neural networks to tackle various problems. In the majority of cases my tool-of-choice Keras library with Tensorflow backend, which has lovely support for multi-GPU training. A lot of my work consists of experimenting with various un-conventional approaches to train deep neural networks, and with a huge amount of experiments comes along quite a lot of failures.

FastAPI and Celery

In this post, I will present to you a simple, minimal working example of utilizing new, high-performance Python web framework FastAPI and Celery - Distributed Task Queue for executing long-running jobs. In addition to the FastAPI framework and Celery distributed task queue, we will also use the RabbitMQ as a messaging queue platform and Redis for returning the results of the executed jobs. For the monitoring of Celery jobs, the [Flower] - Celery monitoring tool will be used.

How to utilize NiaPy for solving KNN parameter optimization problem?

In the following, I’m going to present to you, an example on how to utilize NiaPy micro-framework 1 for solving the parameter optimization problem of K-nearest neighbors classifier 2 against very common Breast Cancer Wisconsin (Diagnostic) Data Set 3. Before we dive in, I suppose you are familiar with the basic understanding of machine learning and the usage of the conventional classifiers such as K-nearest neighbors classifier. If you are only interested in the implementation part you can jump to an Implementation section or view the complete source code on GitHub.