A Credit Assignment Compiler for Joint Prediction
This work addresses the burden of implementing joint prediction methods for machine learning practitioners, though it appears incremental as it builds on existing learning-to-search frameworks.
The paper tackles the challenge of implementing learning-to-search methods for joint prediction tasks by introducing a credit assignment compiler that allows search spaces to be defined via arbitrary imperative programs, resulting in drastically reduced programming and execution time while maintaining high accuracy comparable to alternative approaches.
Many machine learning applications involve jointly predicting multiple mutually dependent output variables. Learning to search is a family of methods where the complex decision problem is cast into a sequence of decisions via a search space. Although these methods have shown promise both in theory and in practice, implementing them has been burdensomely awkward. In this paper, we show the search space can be defined by an arbitrary imperative program, turning learning to search into a credit assignment compiler. Altogether with the algorithmic improvements for the compiler, we radically reduce the complexity of programming and the running time. We demonstrate the feasibility of our approach on multiple joint prediction tasks. In all cases, we obtain accuracies as high as alternative approaches, at drastically reduced execution and programming time.