NEW YORK UNIVERSITY AT TREC 2018 COMPLEX ANSWER RETRIEVAL TRACK

In this paper, we describe our submission to the TREC-CAR 2018. We use a method introduced by Nogueira et al. (2018) to efficiently learn diverse strategies in reinforcement learning for query reformulation and focus minimally on the ranking function. In this framework, an agent consists of multiple specialized sub-agents and a meta-agent that learns to aggregate the answers from sub-agents to produce a final answer. Sub-agents are trained on disjoint partitions of the training data, while the meta-agent is trained on the full training set. Our method makes learning faster, because it is highly parallelizable, and has better generalization performance than strong baselines, such as an ensemble of agents trained on the full data.