Diversifying Dialog Generation via Adaptive Label Smoothing
This addresses the issue of low response diversity in dialogue systems, which is incremental as it builds on existing label smoothing methods by adding adaptability.
The paper tackles the problem of poor diversity in neural dialogue generation caused by over-confidence, proposing an Adaptive Label Smoothing approach that adaptively estimates target distributions per context to improve diversity, with experiments on two benchmark datasets showing it outperforms competitive baselines.
Neural dialogue generation models trained with the one-hot target distribution suffer from the over-confidence issue, which leads to poor generation diversity as widely reported in the literature. Although existing approaches such as label smoothing can alleviate this issue, they fail to adapt to diverse dialog contexts. In this paper, we propose an Adaptive Label Smoothing (AdaLabel) approach that can adaptively estimate a target label distribution at each time step for different contexts. The maximum probability in the predicted distribution is used to modify the soft target distribution produced by a novel light-weight bi-directional decoder module. The resulting target distribution is aware of both previous and future contexts and is adjusted to avoid over-training the dialogue model. Our model can be trained in an end-to-end manner. Extensive experiments on two benchmark datasets show that our approach outperforms various competitive baselines in producing diverse responses.