Biography
2007–2012, studies before PhD: For bachelor degree, I studied mathematical computer science at Masaryk University with focus on algebra and graph theory. It had a significant theoretical element which was fun but it lacked a real application. An enlightening event was passing the course Coding, Cryptography and Cryptographic Protocols, my first A, which I was immediately offered to teach in the consecutive term. I was the youngest person in the class, still without a bachelor’s degree, yet teaching this master’s course. I ended up teaching this course, in English, for 6 academic terms and made it the central focus of my master’s degree. It perfectly utilized my theoretical knowledge in algebra, such as group theory, and applied it in concrete designs of security protocols. As a part of the Erasmus program, I went to Eindhoven University of Technology to work with Berry Schoenmakers. My mentors at Masaryk University were Jozef Gruska and Vaclav Matyas, to whom I am thankful for who I have become.
2012–2018, PhD studies: The security application of cryptographic protocols was fascinating and I wanted to stay in information security systems. I registered for a PhD at Masaryk University in the lab of Pavel Zezula. Our work on gait recognition diverged from cryptography, but still kept the security environment. Gait being a behavioral biometric made me shift towards the domain of pattern recognition. Unfortunately, my lab was merely interested in signal processing and geometric similarity search, while AI was out of its scope. Another influential event was the internship at University of Toronto with Konstantinos Plataniotis, where I got finally introduced to the concepts of machine learning. After returning from Canada, I changed my supervisor to Petr Sojka who was more specialized in AI. We continued on gait recognition and developed various linear models for person identification and re-identification, and a complex evaluation framework. I could not be more obliged to Petr. We published 7 papers together and even got a few awards.
2018–2019, first postdoc: Keeping the security application and advancing the machine learning models, I did my first postdoctoral fellowship at University of South Florida with Sudeep Sarkar. Our main mission was to participate in the TRECVID challenge, called Activity Extended Video, on detecting specific events in long video surveillance footage. Videos of the benchmark VIRAT dataset include streets and parking lots where people and cars move around. In my opinion, this topic is very important to address as it is physically impossible for human security officers to constantly watch multiple security cameras. We designed a couple solutions based on deep neural networks 3DCNN, YOLO, and an end-to-end system of cascaded modules of SSD detector, MOSSE tracker and LSTM recurrent neural net. Our team USF Bulls passed to the second round of the competition and finished sixth of 27. Trying such a competitive challenge was a great experience.
2019–2023, second postdoc: My PhD project on gait recognition has received the Joseph Fourier Prize by Institut Francais and I was kindly asked to come work to France. I saw a great opportunity to apply my skills at the STARS team of Francois Bremond on the topic of computer vision in psychiatric diagnosis. I received the UCA-IDEX thematic postdoctoral grant twice, first with Antitza Dantcheva, who was a recent CRCN, and the second one directly with the director Francois. The seniority of my current position conneced me with PhD students, engineers and interns here at INRIA. At the beginning, I was partly involved in Antitza’s projects ENVISION and RESPECT on face biometrics and lip reading, and later I was fully in the MEPHESTO project focused on analyzing the psychiatric patient-clinician interaction. The project has connected me with foreign computer scientists such as Jan Alexandersson and Philipp Muller at DFKI in Germany and clinicians Philippe Robert and Alexandra Konig here at CoBTek. My contributions were to supervise data collection process and to develop computer vision models for detection of various biomarkers that could help clinicians to analyze symptoms and better diagnose the psychiatric disorders.
2023–present, permanent researcher: I enjoy the collaboration with clinicians because of mutual respect and passion for the topic. This is where I see my future. My first step was to apply for the ERC Starting Grant. ...
© Michal Balazia. All Rights Reserved. Designed by HTML Codex