Prompter: Zero-shot Adaptive Prefixes for Dialogue State Tracking Domain Adaptation
This addresses the problem of adapting DST models to new domains without labeled data for researchers and practitioners in conversational AI, representing a novel application of PETL in zero-shot scenarios.
The paper tackles zero-shot domain adaptation for Dialogue State Tracking by introducing Prompter, a method that uses target domain slot descriptions to generate dynamic prefixes for self-attention layers, achieving state-of-the-art performance on MultiWOZ and SGD benchmarks with improved handling of 'none'-valued slots.
A challenge in the Dialogue State Tracking (DST) field is adapting models to new domains without using any supervised data, zero-shot domain adaptation. Parameter-Efficient Transfer Learning (PETL) has the potential to address this problem due to its robustness. However, it has yet to be applied to the zero-shot scenarios, as it is not clear how to apply it unsupervisedly. Our method, Prompter, uses descriptions of target domain slots to generate dynamic prefixes that are concatenated to the key and values at each layer's self-attention mechanism. This allows for the use of prefix-tuning in zero-shot. Prompter outperforms previous methods on both the MultiWOZ and SGD benchmarks. In generating prefixes, our analyses find that Prompter not only utilizes the semantics of slot descriptions but also how often the slots appear together in conversation. Moreover, Prompter's gains are due to its improved ability to distinguish "none"-valued dialogue slots, compared against baselines.