STRENGTHENING CODE'S AI EXPERTISE:
WELCOME, PROF. NTOUTSI!

As of August 1st, 2022, Prof. Dr. Eirini Ntoutsi has accepted the call to the W3 professorship for Open Source Intelligence at the Department of Computer Science at the Universität der Bundeswehr München. In this interview, she talks about her research, career path and future plans for her new Lab AIML (Artificial Intelligence and Machine Learning) at the RI CODE.

Prof. Ntoutsi, what will you be researching at CODE?

I design intelligent algorithms that learn from data, continuously following the cumulative nature of human learning while ensuring that what has been learned helps drive positive societal impact.

My research currently focuses on three main areas: 1) adaptive learning, that is learning from changing data, e.g., concept drifts, out-of-distribution; 2) responsible AI, that is building AI systems that have a positive impact on society, focusing in particular on fairness-aware learning and transparent and explainable AI and 3) generative AI, that is using machines to generate new plausible data and artifacts.

Why did you choose RI CODE and the UniBw M?

The RI CODE is becoming one of the largest interdisciplinary research centers in cybersecurity and smart data. It brings together a diverse team of experts working on many topics, from network and software security to privacy and disinformation. Artificial Intelligence and Machine Learning methods are nowadays used widely in these areas for, e.g., network monitoring, attack detection, etc., so I believe there are excellent opportunities for developing AIML methods for such complex environments characterized by volume, velocity, variety, and veracity of data.

On the other hand, ML models are vulnerable to attacks, e.g., via training instances; therefore, building resilient methods becomes extremely important. Working towards resilient AI is a direction I want to pursue, and I am confident I can find strong collaborators with complementary expertise at UniBw and RI CODE. I hope through my expertise in AIML, I can contribute to the further development of CODE and UniBw M.

AI is a powerful technology that can help solve the world's biggest challenges, but might also pose risks to individuals and groups. Therefore, we have the responsibility to understand and mitigate its risks and implications. I hope to contribute to the responsible AI direction in research and teaching.

Finally, Munich and Bavaria comprise an excellent place for AI research in Germany and worldwide, with top universities, a strong tech industry, capital availability, and the city's attractiveness.

Where have you done research before and how did you come to your area of expertise?

I became fascinated with Artificial Intelligence during my undergraduate studies at the Computer Engineering and Informatics Department (CEID) at the University of Patras, Greece. In particular, in my diploma thesis, I was training a computer to learn how to play a new strategy game using Reinforcement Learning (RL) and Neural Networks (NNs) and experience through self-played games and human-computer games. Before that, I was a student assistant at the Computer Technology Institute (RACTI) in Patras, where I was extending a Decision Tree induction algorithm for set-valued attributes. My advisor, Prof. Dr. Kalles, pointed me to the "Machine Learning" book by Tom Mitchell, which is still one of my favorite books.

I continued to expand on this topic during my master's studies at the same department, where I focused on text classification.

During my doctoral studies at the University of Piraeus, Greece, I focused on similarity measures between different machine learning models, from decision trees and frequent itemsets to clusters and clusterings, and how model similarity connects to dataset similarity. Such similarity measures are useful to compare models extracted from different datasets, e.g., different branches of a bank or under different parameter settings, for meta-learning, monitoring model changes over evolving data and data streams, etc.

In 2010 I joined the group of Prof. Hans-Peter Kriegel at the Institute for Informatics, Ludwig-Maximilians-Universitaet Munich as a post-doctoral fellow of the Alexander von Humboldt Foundation, working on high-dimensional data streams and model stability.

In 2016 I switched to the Leibniz University Hannover as an associate professor of Intelligent Systems at the Faculty of Electrical Engineering and Computer Science and became affiliated with the L3S Research Center.

In 2021 I joined the Freie Universität Berlin Berlin (FUB) as a full professor of Artificial Intelligence at the Department of Mathematics and Informatics, a position I held until joining UniBw-M and RI CODE.

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

My main goal with the AIML lab is to do cutting-edge research in the field of Artificial Intelligence and Machine learning driven by real-world challenges and to consider the social implications of the technology.

Another goal of the lab is to equip students and young researchers with the necessary skills to build trustworthy and sustainable AI technology.

Regarding our research activities, we will continue working on ongoing research projects like the VW BIAS and the ITN NoBIAS project, which focus on taking into account ethical and legal considerations when designing AI methods. Likewise, we will continue participating in the SFB1463, where we focus on developing machine learning models for evaluating wind-turbine designs and generating novel designs with AI. The same applies to the DFG project Hephaestus, where we focus on designing ML models for adaptive process planning.

At UniBw M, we will start two new EU Horizon 2020 projects, STELAR and MAMMOTH. The STELAR project will build an innovative Knowledge Lake Management System to support and facilitate a holistic approach for FAIR and AI-ready data, with application to the agriculture domain. Our team will focus on tools and techniques for AI-ready data, particularly on data annotation, data interventions, and synthetic data generation. The MAMMOTH project will design and develop tools and techniques for discovering and mitigating discrimination in complex data, including multi-discrimination scenarios and network and multimodal data. Our team will work on fairness-aware learning in the presence of multiple protected attributes like gender, race, and age.

I see a lot of opportunities for joint research projects within CODE and UniBw M, as well as with other institutes from the Munich area, but also nationally and internationally. Examples include xAI for model inspection and debugging, real-time network monitoring, resilient ML, etc. I remain open to new ideas and am always delighted to work with experts from other disciplines. Thus far, I have had the great pleasure of working with experts from different fields, from bioarchaeology and civil engineering to philosophy and law.

Currently, I am looking for multiple Ph.D. students or postdocs on multi-fairness-aware learning, xAI, online fairness-aware learning and adaptive learning to join our group.

What are you most looking forward to?

I look forward to creating a nice new team and working on all these exciting topics. I am lucky that some of my students will join me in this journey either in person or remotely from Berlin and Hannover. And I am looking forward to the new members of the team.

Also, I look forward to connecting with old and new colleagues from the Munich area and being part of the vibrant AI landscape in Munich.

And of course, to enjoy Munich life. Although I have already been living in Munich for the last three years, the commuting and the pandemic didn’t allow me to fully enjoy the city and the wonderful surroundings.

Prof. Dr. Eirini Ntoutsi holds the full professorship of Open Source Intelligence at RI CODE since August 1, 2022. Her research focuses on adaptive learning, responsible AI and generative AI.