[Meet our Faculty] EURECOM welcomes Chiara GALDI as a Professor in the Digital Security department
Q. Could you describe briefly your academic trajectory so far?
CG. I studied Computer Science at the University of Salerno in Italy. During my studies, I had the opportunity to select a few courses on image processing and biometrics. At this point, I also started collaborating with the biometric and image processing group.
Subsequently, I was awarded a scholarship for a PhD at the University of Salerno, and the topic was iris recognition, focusing on biometrics. Shortly after the Ph.D. program started, there was a chance to receive a scholarship from the French embassy in Italy to study in France and complete the joint Ph.D. program there. This is how I ended up for the first time in EURECOM in July 2014. Therefore, I started working on both biometrics and image forensics, combining the two. Finally, I completed my Ph.D. here in France and received a dual-title in France and Italy.
In addition, I attended a master’s program concurrently with the first year of my Ph.D. Since it was a master’s degree in security and intelligent systems, project management, and homeland security. I recall attending the master’s in the morning and working on my PhD in the afternoon. After receiving my PhD in 2016, I was offered a postdoctoral position here at EURECOM. After one year and a half, my contract was changed to Research Associate. Thus, I became a researcher here at EURECOM and stayed until June 1st, 2023, when I got the position of Assistant Professor.
Q. What are the topics of research you have worked on at EURECOM?
CG. Well, I had the opportunity to study different topics. I am inquisitive and have the chance to work on different things I enjoy. At EURECOM, I had to work on various projects, which is the main difference between EURECOM and the University of Salerno and other universities.
I had to work on the multi-biometric systems at the beginning of the project “PROTECT,” which was the first project I attended until the end. This was a vast European project and a lovely experience for me. Then, we also worked on other tasks related to video surveillance, like “OKLOS,” a French ANR project. More specifically, this project examined how to improve video surveillance with image processing and fusion from different spectrums, including the visible and thermal scopes.
More recently, I worked on a project called “XAI-Face” that examined the explainability of facial recognition. That subject is incredibly fascinating. Considering that technology is pervasive, people who work in biometrics can extract data from a person’s face and other biometric characteristics. However, we are unsure of the precise processes taking place right now in these deep neural networks. Therefore, this study aims to aid in the knowledge of how this network functions, what occurs within these networks, and how the information gleaned from faces is used. Eventually, this will help with understanding how to safeguard some unnecessary information.
Q. What novel expertise do you bring to EURECOM?
CG. My background is primarily in biometrics and image forensics. I decided to devote myself to a project dedicated to helping people and human rights. Therefore, I began learning about various subjects, including explainability and, more recently, the problems with bias and discrimination resulting from this new technology. Sometimes, the systems are biased, so they work differently for different categories of people. For example, AI is demonstrated to work better on white males and worse on women and black people. A sort of cross-bias results from the combination of qualities; depending on how the system is designed, there may be more significant prejudice against a group of categories than against others, such as black women. I started looking for a call that accepts project proposals on such themes because I wanted to work on related subjects. Fortunately, several issues were about bias and human rights for the new Horizon calls. I submitted my project proposal on a work addressing discrimination in AI-based technologies in 2022. Although it received a decent grade, it was rejected, but I’ll attempt to submit it again. The goal was to investigate and mitigate the issue of discrimination in AI-based systems. Since I genuinely wanted to work on these topics, I have spent much time creating project proposals lately. But even so, I think it’s a good starting point because it discusses explainability, presumably aiding in comprehending bias and alleviating prejudice and discrimination.
I also included in my brief bio that I wanted to work towards helping and raising awareness for human rights. It can occasionally be challenging for specialists to comprehend what I indicate. A recommendation we made a few years ago was to provide tools for law enforcement agencies to help them better analyze pedophile pornographic content.
So, the subject matter is still biometrics, image processing, and image forensics, but the application is to aid in problem-solving. In other words, real-world examples of application from society.
Q. What do you enjoy at EURECOM?
CG. When you first join EURECOM, the environment, the international aspect, and the sheer number of international students are what you love as a student. It’s fascinating, and it’s a happy place! What convinced me to stay is that while everything is lovely, getting a Ph.D. and doing research are challenging tasks. The fact that there are so many excellent researchers here inspired me to choose to stay. Several scholars receive ERC funding and other awards of a similar nature. I enjoy learning from others, and since there are so many excellent researchers in this field, I would like to take note of what they have accomplished and perhaps emulate it.
Q. If you had to sketch a five-year plan, what would that be as a new assistant professor?
CG. The short-term plan is to keep working on the XAI- face project on explainability and bias for face recognition. Still, I want to bring my perspective based on my background in biometric traits and my multidisciplinary experience in biometrics and image forensics. Sometimes, new ideas can originate from different topics!
Then, I’d like to establish a research interest group focused on these discrimination issues and how to employ computer vision and image forensics for human rights. I want to start my group or subgroup on fairness in computer vision and biometrics, but I’m unsure when. Another thing on the list is to get the HDR title so I can supervise my PhD students. I am contacting the people I met on the job since I started working on my Ph.D. in Italy to build my network and look for people interested in this topic. Researchers sometimes work mainly on their cases with people familiar to them. In some papers, I understood that we still classify people according to their race, using terms like white, black, Latino, etc. Therefore, although they are not directly related, these categories are self-defined by Americans. They might not be entirely correct. Still, the ability to identify people using biometric systems, AI-based systems, and other methods is not very accurate. They are not precise classifications of people. However, experts are working on it from different fields, such as anthropology, and some academics claim that race doesn’t exist. Of course, we use ethnicity as a category rather than race, much like some biometrics researchers do. However, some anthropological professors assert that there is no such thing as ethnicity. Thus, these two terms don’t communicate much with one another. I plan to get professionals working on the legal, ethical, and social aspects of these words together to comprehend them, as well as some knowledge from anthropology and other fields. Since biometrics professionals work with susceptible data, we must be cautious about the categories and classes we use in our systems. We ought to be wiser and more knowledgeable. And I would like to create a multidisciplinary interest group on these topics.
