Jani Bizjak

Name: Jani
Surname: Bizjak
e-mail: jani.bizjak@gmail.com
Phone: +386 1 477 3147
Address: Department of Intelligent Systems
Jožef Stefan Institute
Jamova cesta 39
1000 Ljubljana
jani bizjak

About me

Since I got my first computer I found it intriguing and wanted to know how it works and why. Thirst for knowledge only increased as I got older and for this reason I enrolled into Faculty of Computer and Information science and Faculty of Mathematics and Physics in an interdisciplinary study of mathematics and computer science, both undergraduate and master’s programme.

I chose interdisciplinary study of mathematics and computer science because backbone of most computer programs is mathematics. With extended knowledge of mathematics I gained better understanding of principals behind algorithms and most importantly knew why something works, not only how to use it.

After school I often had some free time which I used to learn something extra about Android systems. I found great interest in developing games and applications for Android devices. I have learned enormous amounts about how to design a game/an application so users will like it, how to implement various algorithms for different tasks and how to use internet to connect people playing/using the game/application.

Ever since enrolling in college I toyed with an idea of going to a foreign country to expand my horizons, to see how education is in different parts of the world and to acquire new connections all over the world from my field of study. In 2013 I got my chance and got Erasmus scholarship for exchange students in Europe. Soon after northern Sweden became my new home.

Sweden experience was incredible. Never have I thought I will love it so much. I met tons of interesting people with a lot of whom I still keep contact. I learned about Swedish culture and explored whole Scandinavia through multiple trips and excursions. I loved the education system, which is much different from Slovenian, more hands on with smaller groups of students per professor. I still can’t believe that most of the professors knew us by our names at the end. There I also got a chance to help at a multimedia conference (MUM 2013) which the faculty was organising. In the end all I can say is that this was probably one of the best experiences of my life.

After coming home from Sweden I got a job offer at “Jožef Stefan” Institute, where I worked as a student/intern before I left for Sweden. After some consideration I accepted the job of Teacher Assistant in Department of Intelligent System, JSI.

My research area is Signal analysis using Deep neural networks. In 2015 I finished my master’s thesis with the title "ECG signal analysis using deep neural networks". The results of the thesis were exceptional as I got nominated for best thesis award (Prešernova nagrada) at University of Ljubljana.

In 2015 I’ve enrolled in a Jožef Stefan International Postgraduate School and am currently working on a PhD thesis with working title: “Human activity recognition using Deep neural networks”.



Title: ECG signal analysis using deep neural networks
Abstract: V tem prispevku je na kratko obrazloženo delovanje srca ter zaznavanje srčnih bolezni s pomočjo elektrokardiografije. Predstavljena je ideja globokih nevronskih mrež, natančneje povratnih nevronskih mrež ter nekaj optimizacijskih metod, ki so nujno potrebne za učenje velikih mrež. V zadnjem delu prispevka so predstavljeni rezultati klasifikacije kompleksa QRS, kjer povratna nevronska mreža doseže 99 odstotno klasifikacijsko točnost.

Title: ECG signal analysis using deep neural networks
Abstract: With rapid increase in computational power, machine learning methods are making huge strides into solving common societal problems. Even in medicine we recently see adaptation of computer systems designed to help medical staff with diagnosis and healthcare support, alerting them on possible risks in patient’s vital signs. In this thesis two deep learning architectures are shown to be able to detect and analyse ECG signals for anomalies caused by heart diseases. First part of the thesis is dedicated to explanation of symptoms for the addressed heart diseases from ECG signal. In second part an introduction to deep neural networks is made, featuring both convolutional and recurrent neural networks. The last part is dedicated to extended analysis of networks performances, where best results show accuracy of over 99 percent.

Title: Object search using mobile platform and RGBD camera
Abstract: Intelligent robots are nowadays making their way from laboratories to domestic homes. In this thesis, a vacuum cleaner robot iRobot Roomba is upgraded so it can search for objects in a room. Robot sensors alone are not good enough that is why a RGBD camera Kinect is added on top of the robot. For software part the robot meta operating system ROS is used. ROS connects the robot’s hardware and software together and is easily upgradable with custom packages for different tasks. This thesis is divided in two parts: object recognition and systematical search of space. For object recognition, a solution made by RoboEarth is used. The focus of this thesis is on navigating the robot and bringing the whole system together. For better representation of the robot’s actions, visualization and verbalization of these actions are added. Before the evaluation of the program on a real robot, exact parameters of the visual recognition system are measured and then used to create a test environment.


Native Slovenian
Full Professional Proficiency English
Limited Working Proficiency Croatian, Serbian
Elementary Proficiency French, Swedish
© 2017 Jani Bizjak. All rights reserved. This material may not be reproduced, displayed, modified or distributed without the express prior written permission of the copyright holder. For permission, contact me .