Hi! My name is Martin Kolarik and this is my personal portfolio and blog web. I am a machine learning researcher based in the enchanting city of Brno in the Czech Republic. You can find my portfolio & CV below. In the blog section I will describe some of our researches in simple human language and write about some non research projects I developed.

Short biography

I was born in 1991 in the town of Znojmo where I attended high school. I choose to go to Brno to graduate from Brno University of Technology in Telecommunications and information technology. After writing my masters thesis in sentiment analysis and NLP text classification, I fell in love with machine learning. My thesis recieved positive reviews and this got me an offer for a Ph.D. programme at the university in the field of machine learning and signal processing under associate professor doc. Ing. Radim Burget, Ph.D. I currently work as a research scientist on projects in applied and fundamental research. My work experiences include both positions in corporate environment and startup companies where I worked during my studies.
I am passionate about new research approaches in the fields of machine learning and data science and I do my best to keep track of current state of the art. In my spare time I enjoy playing the guitar, saxophone and riding & fixing my old motorcycle.

Work experiences

Junior Research Scientist
Brno University of Technology, Oct 2017 – Present
Working as a research scientist on fundamental and applied research projects regarding medical image processing and big data analysis.

Machine learning engineer
BurgSys, a.s., Sep 2018 – Jan 2020
Development of segmentation and novelty detection algorithms for industrial computer vision platform.

Junior developer
Asseco Central Europe, a.s., May 2017 – Apr 2018
Part of the docker delivery team, working with Docker, linux and basic web development support.

Research experiences

TAČR NV18-08-00459, 2018 - 2021
Project focused on developing segmentation algorithms for human spine imaging techniques, developing and 3D printing models of patients spines for pre-operation examination and modelling of gravitational forces on deformed spines.

MPO reg. no. FV20044, with 3S, s.r.o., 2017 - 2020
Development and design of expert system for automatic management of large data repositories using predictive analysis from unlabelled big data.

Education

Ph.D. in machine learning and data science
Brno University of Technology
2017 – 2021 (expected)
My PhD study (ongoing) is focused on efficient learning methods for 3D medical image processing. I am currently focusing on transfer learning for 3D convolutional neural networks and beginning research in parameter efficient equivariant networks for semantic segmentation.

Master's degree (Ing.) in Telecommunications and information technology
Brno University of Technology
2014 – 2017
Title of the Thesis "Deep Learning for Text Classification". Thesis was focused on text sentiment binary classification using convolutional neural networks. Ing. is Czech equivalent of international title MSc.

Bachelor's degree in Telecommunications and information technology
Brno University of Technology
2011 – 2014
Title of the Thesis "Programmable MIDI Controller".

Skills

Key skills:

Machine learning, Python, Java, OOP, Image Processing, Data Science

Tools & Technologies:

Git, SVN, HTML, Docker, Linux, Keras, Tensorflow, Deep Neural Networkss

Publications (selected)

Optimized high resolution 3D dense-U-Net network for brain and spine segmentation
M Kolařík, R Burget, V Uher, K Říha, MK Dutta
Applied Sciences 9 (3), Impact factor 2.217 (2018), Published 2019

Computer vision method for grading of health of a fundus image on basis of presence of red lesions
S Bhattacharya, J Sehgal, A Issac, MK Dutta, R Burget, M Kolarik
2018 41st International Conference on Telecommunications and Signal Processing, Published 2018

Denoise pre-training for segmentation neural networks
M Kolarik
Student EEICT 1, 752-756, Published 2019

For complete list of my publications please visit my Google Scholar profile

Languages

English - C1 (CAE certified)
German - A2
Czech - native speaker

Courses, awards and achievements

1st place in Student EEICT conference in category Signal, data and image processing

I Won first place amongst other Ph.D. students in Student EEICT 2019 conference in the category "Signal, data and image processing" with my paper "Denoise pre-training for segmentation neural networks".

AT&T Mobility and provisioning course

I completed an At&t training center 3-months Mobility and provisioning course in June 2015 focused on linux administration and cellular networks architecture.

Non research projects

NerudAI

Czech poems generating system based on LSTM and Char-rnn. Development of NLP system in localized language was a challenging task. Whole system including printer, user interface design and a thin client based on raspberry pi was a part of Prototyp festival in Brno in 2018 focused on Art&Technology fusion. [LINK]