A Policy for Early Sequence Classification
This addresses the need for efficient real-time predictions in applications like streaming data analysis, though it is incremental as it builds on existing early classification methods.
The paper tackles the problem of early sequence classification by introducing classifier-induced stopping, a supervised approach that improves accuracy and timeliness without waiting for the entire sequence, achieving an average Pareto frontier AUC increase of 11.8%.
Sequences are often not received in their entirety at once, but instead, received incrementally over time, element by element. Early predictions yielding a higher benefit, one aims to classify a sequence as accurately as possible, as soon as possible, without having to wait for the last element. For this early sequence classification, we introduce our novel classifier-induced stopping. While previous methods depend on exploration during training to learn when to stop and classify, ours is a more direct, supervised approach. Our classifier-induced stopping achieves an average Pareto frontier AUC increase of 11.8% over multiple experiments.