National Research Center for Applied Cybersecurity ATHENE

A six-hour outage on the part of Facebook, Instagram and co. translated into a massive loss of revenue for their US-based parent company.

What causes these infrastructure failures? And how can the root cause be detected and resolved as rapidly as possible?

Interactive forensics play a key role when analyzing new, unknown vulnerabilities. This requires large volumes of data to be processed and visualized. In this context, visual analytics methods have proven effective, as they combine automatic data analytics algorithms with new visualization methods, thereby allowing experts to interact with this data.

Modern biometric solutions require a lot of computing power, but digital systems are getting smaller and more compact in today’s world. For biometric recognition systems to work on low-end devices, efficient solutions are needed. What might these efficient solutions look like, without compromising on accuracy and thus security?

Biometric systems are increasingly finding applications in critical areas, such as border control or forensic science, where they can directly affect people’s lives. Unfortunately, studies have consistently shown that biometric systems do not necessarily behave fairly. So why are some biometric models discriminatory, and what is needed to ensure that systems are designed to be fair?

Not every image is equally fit for use in automated face recognition. The suitability of biometric samples for recognition is also referred to as “biometric quality”. Determination of biometric quality is an essential step in the avoidance and prevention of errors. So, how exactly can the quality be established, and what are the factors that influence biometric quality?

The accuracy of biometric systems is driven by current advances in deep learning. However, this progress comes with the inherent disadvantage that the internal relationships of these highly complex models are difficult to understand. But in order to build trust and security in biometric decisions, transparency is essential. So, how can highly complex models be made more explainable and interpretable?

With the passage of time, information about the person depicted in a painting often becomes forgotten. Today, we can only guess which historical person was immortalized by an artist in a painted portrait. How can modern biometric facial recognition help to recover information that was long believed to have been lost?

Time and again, attempts are made to overcome biometric systems. In so-called “presentation attacks,” criminals try to spoof the relevant system with the help of artifices such as face masks, videos or fake images. So, how can such attacks be detected and unauthorized access denied? 

We interviewed Dr. Naser Damer about current research at the ATHENE Center into face morphing.

Biometric Recognition Systems – Benefits and Obstacles in Everyday Consumer Life is a study based on a structured overview of biometric recognition systems currently used by consumers and very likely to be offered for use in the next five years. An online survey was conducted to determine consumer views on biometric recognition systems.

Working on behalf of the Federal Criminal Police Office (Bundeskriminalamt), we evaluated four of the facial identification systems currently available on the market with regard to their recognition performance in forensic investigations.

The ongoing COVID-19 pandemic has raised the profile of hygienic and contactless identity verification. At the same time, however, it has led to widespread wearing of face masks, which are considered essential to keep the transmission of the virus under control. The impact of mask wearing on facial recognition in a collaborative environment is a sensitive but currently neglected topic.