Juergen Schmidhuber is a computer scientist long associated with deep learning research. His homepage describes him as Scientific Director of IDSIA, the Dalle Molle Institute for Artificial Intelligence in Lugano, Switzerland, as well as director of the KAUST AI Initiative at King Abdullah University of Science and Technology and a professor in the Swiss university system.
Schmidhuber is best known as a co-inventor of the Long Short-Term Memory network, or LSTM, the recurrent neural network architecture introduced with his student Sepp Hochreiter that solved the problem of learning over long sequences. His site notes that the lab’s LSTM systems became the first recurrent deep learning networks to win official international competitions, and that LSTM later ran on billions of devices for tasks such as speech recognition and translation.
His homepage also stakes out broader claims about the history of the field, crediting his lab and earlier work with foundational contributions to areas such as self-supervised pre-training, generative adversarial principles, and the attention mechanisms behind modern Transformers. Whatever one makes of the priority debates, his role in recurrent neural networks and LSTM is central to the deep learning era.