This research introduces a novel calibration-gated mechanism using LLM pseudo-observations to solve the cold-start problem in contextual bandits, significant...
Level: advanced
By Maksim Pershin, Ivan Golovanov, Pavel Baltabaev, Natalia Trankova
Category: research