In his publication on LSTM (Long Short-Term Memory), which appeared as early as 1997, Sepp Hochreiter dealt with a component of machine learning that is now indispensable to artificial intelligence. These are recurrent neural networks that can process sequential data and store important information over long periods of time. The three central ‘gates’ – input, forget and output gates – coordinate which information is retained or discarded.
Today, LSTM networks form the basis of numerous applications, from voice assistants and machine translation to autonomous driving and medical data analysis.
After completing his doctorate at the Technical University of Munich, Hochreiter conducted postdoctoral research at the University of Colorado Boulder and the Technical University of Berlin. Today, he is a professor at Johannes Kepler University Linz, where he heads the Institute for Machine Learning and the LIT AI Lab. He also founded the company NXAI, which is developing its own language model based on xLSTM (Extended Long Short-Term Memory), and the Bilateral AI initiative, which promotes exchange between Austrian research institutes and companies in the field of AI.