Sun, March 3, 10:30 AM
90 MINUTESScaling Robot Learning with Offline Data
In recent years, machine learning has achieved remarkable success in various domains, including natural language generation, code synthesis, and photo-realistic image generation, owing to the utilization of high capacity networks and large diverse offline data. Mirroring this trend, robotics is increasingly embracing large-scale datasets and high-capacity networks to advance learning capabilities. In this talk, I will present my recent research on strategies for scaling robot learning with offline data and highlighting key findings and implications for the field.
RL Engineer @ Embark Studios
Ali Ghadirzadeh received his BSc in Electrical Engineering from Ferdowsi University in 2010. He received MSc and PhD degrees from the division of Robotics, Perception and Learning at KTH Royal Institute of Technology in 2013 and 2018, respectively. In 2022, he completed his postdoc assignment at Stanford University on robot learning, the intersection of robotics, computer vision and machine learning. He is currently working as reinforcement learning researcher. His main research interest lies in acquiring robotic skills through physical interaction.