Zero-shot Sequence Labeling: Transferring Knowledge from Sentences to Tokens
This provides a method for evaluating model learning and generating feedback in assistance systems, but it is incremental as it builds on existing visualization techniques.
The paper tackled the problem of inferring token-level labels for binary sequence tagging using networks trained only on sentence-level labels, and found that attention-based methods predicted token labels more accurately than gradient-based methods, sometimes rivaling supervised networks.
Can attention- or gradient-based visualization techniques be used to infer token-level labels for binary sequence tagging problems, using networks trained only on sentence-level labels? We construct a neural network architecture based on soft attention, train it as a binary sentence classifier and evaluate against token-level annotation on four different datasets. Inferring token labels from a network provides a method for quantitatively evaluating what the model is learning, along with generating useful feedback in assistance systems. Our results indicate that attention-based methods are able to predict token-level labels more accurately, compared to gradient-based methods, sometimes even rivaling the supervised oracle network.