On-the-Job Learning with Bayesian Decision Theory
This work addresses the challenge of reducing annotation costs and improving efficiency for deploying machine learning systems in real-time applications, though it is incremental as it builds on existing crowdsourcing and Bayesian methods.
The paper tackles the problem of deploying a high-accuracy system with zero initial training examples by using an on-the-job learning approach that balances real-time crowdsourcing queries with model predictions to reduce costs and improve accuracy over time. On a named-entity recognition task, it achieved more than an order of magnitude cost reduction compared to full human annotation, an 8% F1 improvement over single human labeling, and a 28% F1 improvement over online learning.
Our goal is to deploy a high-accuracy system starting with zero training examples. We consider an "on-the-job" setting, where as inputs arrive, we use real-time crowdsourcing to resolve uncertainty where needed and output our prediction when confident. As the model improves over time, the reliance on crowdsourcing queries decreases. We cast our setting as a stochastic game based on Bayesian decision theory, which allows us to balance latency, cost, and accuracy objectives in a principled way. Computing the optimal policy is intractable, so we develop an approximation based on Monte Carlo Tree Search. We tested our approach on three datasets---named-entity recognition, sentiment classification, and image classification. On the NER task we obtained more than an order of magnitude reduction in cost compared to full human annotation, while boosting performance relative to the expert provided labels. We also achieve a 8% F1 improvement over having a single human label the whole set, and a 28% F1 improvement over online learning.