News Teaching Funding

Perceiving Neural Networks

Perception is a computational feat. The conversion of high-dimensional sensory input to meaning relies on the ability to solve complex pattern recognition problems. Natural tasks like object recognition or visual search are good examples of this process revealing the computational challenges underlying perception. To tackle these challenges complex neural networks have developed in the brain that perform surprisingly well. At the interface between artificial intelligence and neuroscience we focus on uncovering the algorithms and neuro-computational design principles of perceiving neural networks. A practical example of the outcome of this research is a new method for creating artistic images. More specifically, we want to explain how characteristic properties of neural systems originate from the computational requirements of specific perceptual skills.

GitHub

Twitter

Medium

Machine Learning

Deep Neural Networks can challenge humans in complex perceptual tasks. At the same time, neural networks are extremely limited in terms of data efficiency, robustness to minimal perturbations and generalisation to distribution shifts. A major goal of the lab is to shed light on the inner workings of neural networks and to use these insights to rethink the foundations of Deep Learning.

Adversarials

Scene Understanding

Style Transfer

Understanding DNNs

Neuroscience

We are interested in how neural circuits perform the computations that mediate behavior. To tackle this question, we develop machine learning tools, statistical modeling approaches and data analytics solutions for making sense of large-scale high-dimensional neural and behavioral data acquired by our experimental collaborators.

Neural Data Analysis

Markerless Tracking

Human Perception

The human visual system allows us to effortlessly perceive complex properties of the world. While current machine vision systems have made amazing progress on specific tasks in recent years, they do not yet match the generality and robustness of biological vision. We compare how humans and machines see the world in order to better understand both systems.