From machine-learning-based program synthesis to artificial general intelligence
Pocket room, 1st floor
21th November, 12:00-13:00
Machine-learning-based program synthesis is already a reality. The talk will present some results developed in Cluj at the Romanian Institute of Science and Technology, focused on the automated development of web pages. Besides its obvious immediate applications in the information technology industry, program synthesis is also key for the further development of artificial intelligence.
Răzvan Florian is the president of the Romanian Institute of Science and Technology. He is the author of several learning rules for spiking neural networks (the third generation of neural network models). The essential components of one of these rules, which has been developed trough mathematical derivations and computer simulations, have been later found by neuroscientists in the brain. This learning rule allows neurons to learn from rewards and punishments. Components of this learning rule were implemented in hardware in the latest neuromorphic chip developed by Intel. Dr. Florian has also founded two companies. He is now working on creating intelligent systems that are able to autonomously develop software.