The challenges for machine learning

Pascal Schöttle is Professor for IT Security and Machine Learning at the Department of Digital Business & Software Engineering at MCI | The Entrepreneurial School in Innsbruck. Our expert recently answered questions for top.tirol magazine. In the interview, Pascal talks about the challenges facing machine learning, particularly in the areas of security and data protection. Here is a brief outline:

Machine learning can be manipulated by deliberate deception, for example glasses. Algorithms then make false identifications. This vulnerability affects facial recognition, among other things. More and more areas such as payment systems, autonomous vehicles and security systems are dependent on machine learning and could potentially be attacked. It is difficult to effectively protect against such attacks without compromising the efficiency of machine learning systems. Research focuses on making attacks as elaborate as possible to reduce potential attackers. Nevertheless, such attacks are expected to occur on an increasing scale.

You can find the interview in full length online.