Bild: © UniBw M


German version >>

As of October 1st, 2023, Prof. Dr. Marta Gomez-Barrero has accepted the call to the W3 professorship for Machine Learning at the Department of Computer Science at the University of the Bundeswehr Munich (UniBw M). In this interview, she talks about her research, career path and future plans for her new Lab BioML (Biometrics and Machine Learning) at the Research Institute CODE (RI CODE).


Professor Gomez-Barrero, where have you done research before and how did you come to your area of expertise?

"It was in the last year of my Master studies at Universidad Autónoma de Madrid, in Spain, that I first got in touch with IT-security during an internship in a research lab, the ATVS. I had the chance of implementing attacks on biometric systems to reverse engineer the templates stored in the database and thus have access to reconstructions of the original biometric samples or images (e.g., face and iris images or handwritten signatures). The research group also gave me the opportunity of submitting and then presenting my work in a European workshop in Brandenburg in 2011. This is when I decided to pursue a PhD in the broader topic of security and privacy aspects of biometric systems, with internships in research labs at the Hochschule Darmstadt, the NTNU in Gjøvik, and the Università Roma Tre. This allowed me to get in touch with other researchers and learn other languages, my second intellectual passion.

Right after concluding my PhD studies at Universidad Autónoma de Madrid in 2016, I moved to Darmstadt for a PostDoc position at the da/sec research group of the Hochschule Darmstadt, within the National Research Center for Applied Cybersecurity (ATHENE). During almost four years I could expand my network, as well as gain experience in acquiring research projects and mentoring PhD students. In 2020 I moved to Ansbach for a Professorship on IT-security and technical data privacy, my last role before joining the UniBw M.

During this time, the main topic of my research has been biometric recognition. Not tied to a specific modality, such as face or fingerprint, but in a more comprehensive manner, developing general methods applicable to more or less any biometric characteristic, under constrained (e.g., an office with fixed illumination) and non-constrained (e.g., outside on the street) scenarios. I started developing attacks to biometric systems to then focus on developing privacy-friendly systems in compliance with the corresponding ISO standards. Such systems can also take several biometric characteristics (e.g., face and iris) as input in order to provide a more accurate and robust decision. In parallel to those privacy aspects, I have worked on the development of presentation attack detection algorithms to increase the security of biometric recognition schemes. All of this trying at the same time to minimise potential recognition accuracy losses. This involves a deep knowledge of pattern recognition algorithms, mostly machine and deep learning based, and of cryptographic techniques."

What will you be researching at CODE?

"Whereas the topics I have explored so far still pose challenges and offer room for improvements, in the last few years other issues have gained on visibility: why are biometric systems biased towards one particular demographic group? Deep learning algorithms achieve lower error rates, but they seem to operate in a black-box way… how can we explain the decision-making process and the final results? In addition, newer and better encryption schemes, either robust to quantum computing or more efficient homomorphic encryption schemes, have also found their way into the state of the art. Nevertheless, their application to machine learning approaches is still a hot research topic. These will thus be the main topics to be explored at CODE in the coming years: I plan on starting with explainable machine and deep learning to tackle for instance bias in biometric systems, the use of differential privacy and federated learning to increase the security and privacy of biometric systems, and at a later stage I would also research the use of quantum computing for machine learning."

Why did you choose RI CODE and the University of the Bundeswehr Munich?

"In my opinion, the deployment of reliable, fair, secure authentication mechanisms is key to increase safety and security in our society, especially in difficult moments like the ones we are now living. And not only to log in more comfortably into your laptop, but also for access or border control. Working together with law enforcement authorities (e.g., BKA or BLKA, Bundeswehr, BSI, and BND) can only facilitate this process. And the UniBw M is probably the best place to be for such a cooperation.

I am also glad that cyber security is gaining in importance from the perspective of policy makers. This is most likely what has allowed CODE to rapidly expand and embrace different areas within the wide landscape of IT security. Such interdisciplinarity allows for an optimal working environment, where challenges can be discussed with colleagues having different but somewhat overlapping areas of expertise. So are brilliant ideas born, which in turn leads to a positive advance in our society.

And not to be forgotten are the top IT equipment and infrastructure provided by the university and CODE. Without this, no top research can be carried out."

What is your vision for building your new research group at RI CODE?

"My main goal with the BioML research lab is to work on hot, practical research topics in the wider area of machine learning and its applications to biometric recognition and IT-security. This can be best done by combining hands-on teaching and supervision of student-led research projects.

I am already in touch with some academic colleagues in neighboring (Austria, Czech Republic) and more distant countries (USA, Brasil) in order to acquire further research projects. Some commercial contacts have also expressed their interest in collaborating with federal agencies, and I am sure that moving to the Munich area will also increase the number of industrial partners in my projects. Thus I am positive about the rapid growth of the new team.

And last but not least, I am currently looking for multiple PhD students or postdocs who want to join me on this exciting journey! Please get in touch!"

What are you most looking forward to?

"I am looking forward to starting a new team in biometric recognition and machine learning in Munich, within the CODE research institute. Having the support of all the new colleagues, with deep expertise in different but complementing aspects of IT-security, and having the chance of enjoying the vibrant life of Munich and Bavaria, is a sure recipe for success."

Prof. Dr. Marta Gomez-Barrero holds the full professorship of Machine Learning at RI CODE since October 1st, 2023. Her research focuses on biometric systems and privacy-friendly authentication systems.

Please also note the current job offers of the research group on the CODE career portal:

Image: © University of the Bundeswehr Munich