Contextual Bandits Evolving Over Finite Time
This work addresses a specific problem in sequential decision-making for researchers in bandit algorithms, offering a solution to biases in existing policies.
The paper tackles the challenge of contextual bandits with decaying positive externalities, where early mistakes are hard to correct, and proposes a rejection-based policy that achieves low regret independent of the reward matrix structure.
Contextual bandits have the same exploration-exploitation trade-off as standard multi-armed bandits. On adding positive externalities that decay with time, this problem becomes much more difficult as wrong decisions at the start are hard to recover from. We explore existing policies in this setting and highlight their biases towards the inherent reward matrix. We propose a rejection based policy that achieves a low regret irrespective of the structure of the reward probability matrix.