TY  - RPRT
TI  - Autodiscover: A reinforcement learning recommendation system for the cold-start imbalance challenge in active learning, powered by graph-aware thompson sampling
AU  - Parsa Vares
PY  - 2026
UR  - https://arxiv.org/abs/2602.05087
ID  - 2602.05087
ER  -