Inaugural Lectures - Prof. Caglar Gulcehre and Prof. Nicolas Flammarion
Event details
| Date | 12.03.2024 |
| Hour | 18:00 › 19:30 |
| Speaker | Prof. Caglar Gulcehre, Prof. Nicolas Flammarion |
| Location | |
| Category | Inaugural lectures - Honorary Lecture |
| Event Language | English |
Date: Tuesday 12 March 2024
Program:
Registration: Click here
****************************************************************
Prof. Caglar Gulcehre
Bridging Generative AI and Reinforcement Learning Towards a Safer and Brighter Future
Abstract
Generative AI algorithms and foundation models such as ChatGPT are having an all-encompassing impact on society and science. Focusing on the imperative of building safe and efficient AI models, this lecture will unveil the synergies between reinforcement learning and generative AI approaches, presenting opportunities for enhanced adaptability and ethical considerations to enable positive impacts on essential and challenging applications such as AI for science, and robotics. In this talk, I will introduce generative AI and reinforcement learning algorithms with a concise exploration of cutting-edge developments – shedding light on the transformative potential of bridging generative AI and reinforcement learning for responsible and impactful AI systems. I will conclude the talk with interesting future challenges in AI research with a positive outlook into what might be possible in the future with safe AI integration.
About the speaker
Caglar Gulcehre is an assistant professor at EPFL, leading the CLAIRE lab. His research revolves around building intelligent agents through the lens of efficiency, safety, and robustness for challenging real-world environments. He is motivated by solving real-world problems that would have a positive societal impact, such as the applications of AI for science and robotics. He finished his PhD under Yoshua Bengio at Mila (Quebec AI Institute). He previously worked at Google DeepMind, Microsoft Research, and IBM Research on AI and Machine Learning. After more than 6 years at DeepMind as a staff research scientist, he recently transitioned into becoming a professor at EPFL. His work has been published in journals such as Nature, JMLR, TMLR, and Neurocomputing and at conferences like ICML, NeurIPS, ICLR, AISTATS, ACL, and EMNLP. He has won the Best Paper Award at NeurIPS and the Best Paper runner-up at ICML. His work has been featured in several media outlets such as Verge, MIT News, BBC, Forbes, and the New York Times.
****************************************************************
Prof. Nicolas Flammarion
Follow the Gradient: A Tour of Neural Network Theory
Abstract
With ChatGPT and the latest advances in Large Language Models, artificial intelligence is the talk of the town. However, the theoretical foundations of such large machine learning models remain unclear. In this talk, I will discuss recent results that shed light on one of the mysteries behind this success: why gradient methods converge to models that generalize well. We will begin our journey by discussing simple linear regression, then move on to explore one-hidden-layer neural networks, and finally, investigate similar behavior in deep neural networks.
About the speaker
Nicolas Flammarion is a tenure-track assistant professor in computer science at EPFL. Prior to that, he was a postdoctoral fellow at UC Berkeley, hosted by Michael I. Jordan. He received his PhD in 2017 from Ecole Normale Supérieure in Paris, where he was advised by Alexandre d’Aspremont and Francis Bach. In 2018 he received the Fondation Mathématique Jacques Hadamard prize for the best PhD thesis in the field of optimization and in 2021 a NeurIPS Outstanding Paper Award. His research focuses primarily on learning problems at the interface of machine learning, statistics and optimization.
Program:
- 18:00-18:05: Introduction by Prof. Rüdiger Urbanke, Dean of the IC School
- 18:05-18:35: Inaugural Lecture Prof. Caglar Gulcehre
- 18:35-18:45: Q & A
- 18:45-18:50: Introduction by Prof. Rüdiger Urbanke, Dean of the IC School
- 18:50-19:20: Inaugural Lecture Prof. Nicolas Flammarion
- 19:20-19:30: Q & A
- 19:30-21:00: Apéritif in the FoodLab Alpine restaurant
Registration: Click here
****************************************************************
Prof. Caglar Gulcehre
Bridging Generative AI and Reinforcement Learning Towards a Safer and Brighter Future
Abstract
Generative AI algorithms and foundation models such as ChatGPT are having an all-encompassing impact on society and science. Focusing on the imperative of building safe and efficient AI models, this lecture will unveil the synergies between reinforcement learning and generative AI approaches, presenting opportunities for enhanced adaptability and ethical considerations to enable positive impacts on essential and challenging applications such as AI for science, and robotics. In this talk, I will introduce generative AI and reinforcement learning algorithms with a concise exploration of cutting-edge developments – shedding light on the transformative potential of bridging generative AI and reinforcement learning for responsible and impactful AI systems. I will conclude the talk with interesting future challenges in AI research with a positive outlook into what might be possible in the future with safe AI integration.
About the speaker
Caglar Gulcehre is an assistant professor at EPFL, leading the CLAIRE lab. His research revolves around building intelligent agents through the lens of efficiency, safety, and robustness for challenging real-world environments. He is motivated by solving real-world problems that would have a positive societal impact, such as the applications of AI for science and robotics. He finished his PhD under Yoshua Bengio at Mila (Quebec AI Institute). He previously worked at Google DeepMind, Microsoft Research, and IBM Research on AI and Machine Learning. After more than 6 years at DeepMind as a staff research scientist, he recently transitioned into becoming a professor at EPFL. His work has been published in journals such as Nature, JMLR, TMLR, and Neurocomputing and at conferences like ICML, NeurIPS, ICLR, AISTATS, ACL, and EMNLP. He has won the Best Paper Award at NeurIPS and the Best Paper runner-up at ICML. His work has been featured in several media outlets such as Verge, MIT News, BBC, Forbes, and the New York Times.
****************************************************************
Prof. Nicolas Flammarion
Follow the Gradient: A Tour of Neural Network Theory
Abstract
With ChatGPT and the latest advances in Large Language Models, artificial intelligence is the talk of the town. However, the theoretical foundations of such large machine learning models remain unclear. In this talk, I will discuss recent results that shed light on one of the mysteries behind this success: why gradient methods converge to models that generalize well. We will begin our journey by discussing simple linear regression, then move on to explore one-hidden-layer neural networks, and finally, investigate similar behavior in deep neural networks.
About the speaker
Nicolas Flammarion is a tenure-track assistant professor in computer science at EPFL. Prior to that, he was a postdoctoral fellow at UC Berkeley, hosted by Michael I. Jordan. He received his PhD in 2017 from Ecole Normale Supérieure in Paris, where he was advised by Alexandre d’Aspremont and Francis Bach. In 2018 he received the Fondation Mathématique Jacques Hadamard prize for the best PhD thesis in the field of optimization and in 2021 a NeurIPS Outstanding Paper Award. His research focuses primarily on learning problems at the interface of machine learning, statistics and optimization.
Practical information
- Informed public
- Registration required