1971–, Computer scientist
Sven Behnke is a German computer scientist whose 2003 thesis and subsequent papers introduced and demonstrated GPU-trained convolutional neural networks several years before AlexNet's 2012 ImageNet result. His group's CNNs achieved competitive results on handwriting recognition and traffic-sign recognition in the late 2000s, demonstrating that GPU acceleration could make deep CNN training practical.
The work is somewhat overshadowed by the AlexNet result that consolidated the deep-learning shift, but represents an important precursor. Behnke holds a chair at the University of Bonn and continues research in robotic perception and humanoid robotics.
Related people: Alex Krizhevsky, Yann LeCun
Discussed in:
- Chapter 11: CNNs, CNNs in Vision