Full professor Artificial Intelligence Tilburg University
Keynote: The unreasonable effectiveness of overparameterization
9:00 – 10:00
Abstract
The recent developments in deep learning and generative AI have led to many breakthroughs across many application domains. In the presentation, a brief review of the history of artificial intelligence will be followed by an explanation of the basic principles of deep learning and generative AI. It will be shown how a departure from the scientific adagio to keep models as simple as possible (known as “Occam’s razor“) led to the current AI revolution characterized by highly successful neural network models with far too many parameters. The similarities and dissimilarities of artificial intelligence and natural intelligence will be outlined to debunk many false claims about artificial ihtelligence. Finally, the current developments will be described in a down-to-earth way to facilitate a hype-resistant assessment of near-future developments.
Bio Eric Postma
Eric Postma is a full professor Artificial Intelligence at the Cognitive Science & AI department of Tilburg University and at the Jheronimus Academy of Data Science in ‘s-Hertogenbosch of Tilburg University and Eindhoven University of Technology. Eric has worked on neural networks throughout his career. His research tries to unravel the pattern recognition abilities of machines by studying the latest neural architectures. He has worked on computer vision and signal processing tasks across many application domains such as visual arts, health, sustainable development goals, astronomy, and human behavior.