Mengjiao Yang

Papers

Total Citations

H-Index

About

Mengjiao Yang is a machine learning researcher whose work sits at the intersection of reinforcement learning, offline evaluation, and decision-making systems. Her most recognized contribution to the field is her work on benchmarking methodologies for deep off-policy evaluation (OPE), a critical area that addresses how complex policies can be rigorously assessed using large offline datasets — without requiring costly or risky real-world experimentation. This research tackles a fundamental challenge in deploying reinforcement learning to high-stakes domains such as healthcare, robotics, and recommender systems, where live policy testing may be impractical or dangerous. Her 2021 paper, "Benchmarks for Deep Off-Policy Evaluation," has garnered 25 citations, reflecting growing community interest in establishing reliable, standardized tools for offline reinforcement learning research. By providing structured benchmarks, Yang's work empowers researchers and practitioners to more confidently evaluate and select policies in real-world settings — a contribution with meaningful implications for the responsible deployment of AI systems. Her research addresses a gap between theoretical advances in reinforcement learning and the practical demands of safe, data-driven decision making, making her a valuable voice in the evolving landscape of applied machine learning.