|
I am a fourth-year PhD student in Computer Science at University of Southern California. I am lucky to be advised by Prof. Erdem Bıyık. My Research interests lie in Robot Learning, Human-in-the-loop Learning, and Reinforcement Learning. Specifically, I am interested in developing methods that enable robots to continuously improve after deployment by efficiently learning from interaction and feedback in the real world. Toward this goal, I study how diverse forms of human feedback—including demonstrations, interventions, and preferences—can be leveraged for sample-efficient policy adaptation and fine-tuning. I am also interested in developing learning frameworks that make real-world reinforcement learning more practical and scalable. Ultimately, my goal is to enable general-purpose robots that can continually improve their capabilities while adapting to human goals, preferences, and changing environments. Previously, I completed my M.S. at UC San Diego, where I worked with Prof. Xiaolong Wang. I have also spent summers interning at Toyota Research Institute, and [Google] Intrinsic. Fun Fact: My name is pronounced as Yeet. |
|
|
PhD. in Computer Science, University of Southern California
2023-present
|
|
Jul 2026 - Organizing Roadmap to Sample-Efficient Real-World Reinforcement Learning Workshop at CoRL 2026!
|
|
|
|
|
|
Yigit Korkmaz code Transformer networks are highly utilized in Natural Language Processing(NLP) tasks. In this project, the use of Longformer, which is a transformer based network with advanced attention mechanism, in classifying argumentative elements of students’ writing will be discussed. |
|
Yigit Korkmaz code In this project, APF(artificial potential fields), which a method used to formulate obstacle and goal interactions in a robot’s path, is implemented. With the assumption of a known map, obstacles act as repulsive forces where targets act as attractive ones. The code is written for ROS environment, and experimented on real robots. |
|
Template is from Jon Barron's awesome website. |