Repulsive Population Policy Gradients for Behaviorally Diverse Reinforcement Learning
Saeed Hedayatian, Harsh Shah, Celine Mneimneh, and Stefanos Nikolaidis.
arXiv preprint arXiv:2610.36425, 2026.
PhD candidate · Computer Science · University of Southern California
I am a PhD candidate in Computer Science at the University of Southern California. I am broadly interested in sequential decision-making problems. I actively look for sources of inspiration in adjacent fields like generative modeling and quality-diversity optimization and explore various applications to open-ended learning, multi-agent systems, and meta-learning.
I am advised by Prof. Stefanos Nikolaidis and am part of the ICAROS lab at USC.
Prior to starting my PhD, I got my bachelor’s degree in Computer Science from Sharif University of Technology, Iran.
Outside of research, I enjoy reading of all sorts, from philosophy to psychoanalysis to physics! I’m also an amateur cinephile, check out my letterboxd!
Research interestsReinforcement learning, quality-diversity optimization, generative modeling, open-ended learning.
Saeed Hedayatian, Harsh Shah, Celine Mneimneh, and Stefanos Nikolaidis.
arXiv preprint arXiv:2610.36425, 2026.
Saeed Hedayatian and Stefanos Nikolaidis.
International Conference on Learning Representations (ICLR), 2026.
Saeed Hedayatian and Stefanos Nikolaidis.
International Conference on Learning Representations (ICLR), 2026.
Saeed Hedayatian*, Varun Bhatt*, Bryon Tjanaka, Michael Lewis, Katia P. Sycara, and Stefanos Nikolaidis.
ToM4AI Workshop at AAAI, 2026.
Seyed Roozbeh Razavi Rohani, Saeed Hedayatian, and Mahdieh Soleymani Baghshah.
IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2022.
NeurIPS and ICLR.
USC and Sharif University of Technology.
Courses in algorithms, theory of computing, discrete mathematics, AI, NLP, automata theory, and image processing.
The Busy Beaver function and the limits of computability.
Mathematics, Comma Institute for High School Students, Tehran.
Combinatorics, graph theory, and algorithms for olympiad-track students.