Computer Science Seminar: Victoria Dean (Olin College of Engineering)

February 10, 2025 | 11:00am - 12:00pm

Milstein 402 & Zoom

Add to Calendar 2025-02-10 11:00:00 2025-02-10 12:00:00 Computer Science Seminar: Victoria Dean (Olin College of Engineering) Speaker: Victoria Dean (Olin College of Engineering) Title: From Reinforcement Learning to Robot Learning: Leveraging Prior Data and Shared Evaluation The seminar will be available for in-person and Zoom participation. To participate online, please email inquiry-cs@barnard.edu to receive the Zoom link. Unlike most machine learning applications, robotics involves physical constraints that make off-the-shelf learning challenging. Difficulties in large-scale data collection and training present a major roadblock to applying today’s data-intensive algorithms. Robot learning has an additional roadblock in evaluation: every physical space is different, making results across labs inconsistent. Two common assumptions of the robot learning paradigm limit data efficiency. First, an agent typically assumes isolated environments and no prior knowledge or experience – learning is done tabula-rasa. Second, agents typically receive only image observations as input, relying on vision alone to learn tasks. In this talk, I will present work that lifts these two assumptions, improving the data efficiency of robot learning by leveraging multimodality (images and audio) as well as pretraining. I will then introduce a real-world manipulation benchmark for evaluating the generalization of both visual and policy learning methods via shared data and hardware. Finally, I will discuss ongoing work leveraging prior data and benchmarking in another robot learning application: building energy optimization. Victoria Dean is an Assistant Professor of Computer Science at Olin College of Engineering. She completed her PhD in Robotics at Carnegie Mellon University in 2023, where she was an NSF Graduate Research Fellow and a Siebel Scholar. She works on reinforcement learning applications in robotics and building energy optimization. She is passionate about teaching and mentoring undergraduate students, with an emphasis on improving diversity, equity, and inclusion. Outside of computer science, she enjoys experimental baking, swing dancing, and knitting. Milstein 402 & Zoom Barnard College barnard-admin@digitalpulp.com America/New_York public
Campus Map

Speaker: Victoria Dean (Olin College of Engineering)

Title: From Reinforcement Learning to Robot Learning: Leveraging Prior Data and Shared Evaluation

The seminar will be available for in-person and Zoom participation. To participate online, please email inquiry-cs@barnard.edu to receive the Zoom link.

Unlike most machine learning applications, robotics involves physical constraints that make off-the-shelf learning challenging. Difficulties in large-scale data collection and training present a major roadblock to applying today’s data-intensive algorithms. Robot learning has an additional roadblock in evaluation: every physical space is different, making results across labs inconsistent.

Two common assumptions of the robot learning paradigm limit data efficiency. First, an agent typically assumes isolated environments and no prior knowledge or experience – learning is done tabula-rasa. Second, agents typically receive only image observations as input, relying on vision alone to learn tasks. In this talk, I will present work that lifts these two assumptions, improving the data efficiency of robot learning by leveraging multimodality (images and audio) as well as pretraining. I will then introduce a real-world manipulation benchmark for evaluating the generalization of both visual and policy learning methods via shared data and hardware. Finally, I will discuss ongoing work leveraging prior data and benchmarking in another robot learning application: building energy optimization.

Victoria Dean is an Assistant Professor of Computer Science at Olin College of Engineering. She completed her PhD in Robotics at Carnegie Mellon University in 2023, where she was an NSF Graduate Research Fellow and a Siebel Scholar. She works on reinforcement learning applications in robotics and building energy optimization. She is passionate about teaching and mentoring undergraduate students, with an emphasis on improving diversity, equity, and inclusion. Outside of computer science, she enjoys experimental baking, swing dancing, and knitting.

Computer Science