Xerrada Sony AI
- https://telecos.upc.edu/ca/esdeveniments/xerrada-sony-ai
- Xerrada Sony AI
- 2023-05-15T12:15:00+02:00
- 2023-05-15T13:30:00+02:00
Quan?
15/05/2023 de 12:15 a 13:30 (Europe/Madrid / UTC200)
On?
C4002
Afegiu l'esdeveniment al calendari
The talk revolves around two main topics, each presented by the respective speaker: AI in videogames and AI in gastronomy:
The game AI project uses Reinforcement learning to train gameplaying AI agents for challenging PlayStation titles. In this vein GT Sophy was created, an AI capable of winning against human professionals at the game of Gran Turismo. A tuned version of Sophy was recently released to the wider public to popular acclaim. In this talk Daniel will present the inner workings of Sophy and it's training procedure.
In Sony AI's Gastronomy Flagship project they explore what role(s) data-driven AI systems can play in supporting innovation and chef’s creativity, as well as in enriching the overall dining experience. This includes explorations into data-based flavour pairing and recipe generation, the development of different representation approaches for sensory data relevant for gastronomy use cases, and research into the AI-guided creation of multisensory dining experiences.
The game AI project uses Reinforcement learning to train gameplaying AI agents for challenging PlayStation titles. In this vein GT Sophy was created, an AI capable of winning against human professionals at the game of Gran Turismo. A tuned version of Sophy was recently released to the wider public to popular acclaim. In this talk Daniel will present the inner workings of Sophy and it's training procedure.
In Sony AI's Gastronomy Flagship project they explore what role(s) data-driven AI systems can play in supporting innovation and chef’s creativity, as well as in enriching the overall dining experience. This includes explorations into data-based flavour pairing and recipe generation, the development of different representation approaches for sensory data relevant for gastronomy use cases, and research into the AI-guided creation of multisensory dining experiences.
The talk will be divided in two blocks of 30 minutes per researcher with extra time at the end of the talk for Q&A.
Comparteix: