Learning to Decode for Future Success
Learning to Decode for Future Success
复制标题
DOI：
--
复制
发表时间：
2017-01
期刊：
影响因子：
--
通讯作者：
Jiwei Li;Will Monroe;Dan Jurafsky
中科院分区：
文献类型：
--
作者：
Jiwei Li;Will Monroe;Dan Jurafsky
我们引入了一种简单的通用策略来操纵神经解码器的行为，使其能够生成具有特定属性的输出（例如，预先指定长度的序列）。该模型可以被认为是参与者-评论家模型的简单版本，该模型使用参与者（基于mle的令牌生成策略）和评论家（估计所需属性的未来值的值函数）的插值来进行决策。我们证明，该方法能够结合标准神经序列解码器无法处理的各种属性，例如序列长度和向后概率（给定目标的源的概率），此外，当要优化的属性是BLEU或ROUGE分数时，该方法还能在抽象摘要和机器翻译方面产生一致的改进。
We introduce a simple, general strategy to manipulate the behavior of a neural decoder that enables it to generate outputs that have specific properties of interest (e.g., sequences of a pre-specified length). The model can be thought of as a simple version of the actor-critic model that uses an interpolation of the actor (the MLE-based token generation policy) and the critic (a value function that estimates the future values of the desired property) for decision making. We demonstrate that the approach is able to incorporate a variety of properties that cannot be handled by standard neural sequence decoders, such as sequence length and backward probability (probability of sources given targets), in addition to yielding consistent improvements in abstractive summarization and machine translation when the property to be optimized is BLEU or ROUGE scores.