Springe direkt zu Inhalt

Lorenz Linhardt

Photographer: Felix Noak

Photographer: Felix Noak

BUA Institution

Technische Universität Berlin

Description of Research

Understanding what deep neural networks have learned, how this relates to human concepts, and how to manipulate/fix their internal representations. This includes analyses of neural networks using human behavioural data, as well as interventions on their internals using techniques from explainable/interpretable machine learning.

Where in the world has your career been largely based until now?

The D-A-CH region; and I have enjoyed my stay in each of its countries.

Why Berlin?

Berlin offers a vibrant research community, with many academics and other experts to learn from - in particular in machine learning. I also appreciate that there are multiple excellent research institutions seeking to complement their individual expertises in collaborative projects.
Besides that, I enjoy the liveliness of the city and quality of life in Berlin, as well as the friendly and stimulating environment that the machine learning group offers.

What fascinates you about your research area?

One of my favourite aspects of machine learning is that "you get to play in everyone's backyard" (a phrase attributed to the Statistician John Tukey, but equally valid for ML). For me, this has led to very stimulating collaborations across e.g. astronomy, cyber security, and legal studies.
More recently, it has become fascinating for me to closely follow the rapid development of AI capabilities, as well as attempts to understand the underlying mechanisms.

How did you become interested in your specific topic?

I have been interested in various, seemingly disconnected topics in machine learning and figured that learning more about the internals of neural networks (e.g. using explainability techniques or representation analyses) would be the fundamental to all of them.

What did you want to be when you grow up?

As a child, I wanted to become a retiree. I have since decided to add a few steps before that and am not unhappy about this decision.

If you could have a radical career change for a week, what would it be?

At the moment, pathologist, as I would be very interested in understanding first-hand how they work and what they pay attention to when making a diagnosis.