Shubham Bharti

Papers

Total Citations

H-Index

About

Shubham Bharti’s research lies at the intersection of reinforcement learning, machine teaching, and sample complexity theory, with a focus on how intelligent agents can be efficiently guided through reward-based interactions. His most-cited work, “The Sample Complexity of Teaching by Reinforcement on Q-Learning” (2021, 3 citations), introduces a rigorous framework for quantifying the teaching dimension in reinforcement learning—a paradigm where a teacher shapes a student’s policy through carefully designed rewards rather than explicit demonstrations. This contribution is foundational for understanding the minimal number of interactions needed to teach an agent a target behavior, with implications for robotics, adaptive tutoring systems, and human-AI collaboration. By formalizing the sample complexity of teaching in Q-learning, Bharti bridges gaps between theoretical machine learning and practical algorithm design. His work is particularly notable for distinguishing teaching-by-reinforcement from teaching-by-demonstration, offering a more scalable approach for dynamic environments. As an emerging voice in the field, Bharti’s research continues to influence how we think about efficient, reward-driven knowledge transfer in autonomous systems.