A Mutual Information Maximization Approach for the Spurious Solution Problem in Weakly Supervised Question Answering
This addresses a key challenge in weakly supervised QA, improving model reliability by reducing reliance on spurious solutions, though it is an incremental advance over prior heuristic-based methods.
The paper tackles the spurious solution problem in weakly supervised question answering, where models may learn incorrect reasoning paths that coincidentally yield correct answers, by proposing a method that maximizes mutual information between question-answer pairs and predicted solutions, resulting in significant performance improvements on four datasets.
Weakly supervised question answering usually has only the final answers as supervision signals while the correct solutions to derive the answers are not provided. This setting gives rise to the spurious solution problem: there may exist many spurious solutions that coincidentally derive the correct answer, but training on such solutions can hurt model performance (e.g., producing wrong solutions or answers). For example, for discrete reasoning tasks as on DROP, there may exist many equations to derive a numeric answer, and typically only one of them is correct. Previous learning methods mostly filter out spurious solutions with heuristics or using model confidence, but do not explicitly exploit the semantic correlations between a question and its solution. In this paper, to alleviate the spurious solution problem, we propose to explicitly exploit such semantic correlations by maximizing the mutual information between question-answer pairs and predicted solutions. Extensive experiments on four question answering datasets show that our method significantly outperforms previous learning methods in terms of task performance and is more effective in training models to produce correct solutions.