ctc和attention-based解码器的比较

Attention是一种权重向量或矩阵，其往往用在Encoder-Decoder架构中，其权重越大，表示的context对输出越重要。计算方式有很多亚种，但是核心都是通过神经网络学习而得到对应的权重。

Attention模型虽然好，但是还是有自身的问题[4][5]，问题如下：

1）适合短语识别，对长句子识别比较差

2）noisy data的时候训练不稳定

因此比较好的方法是使得Attention与CTC进行结合，对比Attention模型还有CTC模型，Attention+CTC模型更快的收敛了，这得益于初始阶段CTC的阶段对齐更准确，使得Attention模型训练收敛更快。----参考：https://blog.csdn.net/quhediegooo/article/details/76842201

 

The attention model has often been shown to improve the performance over another end-to-end approach, the Connectionist Temporal Classification CTC), mainly because it explicitly uses the history of the target character without any conditional independence assumptions. 但是the attention model has shown poor results especially in noisy condition and is hard to be trained in the initial training stage with long input sequences,as compared with CTC. 原因This is because the attention model is too flexible to predict proper alignments in such cases due to the lack of left-to right constraints as used in CTC.结合他们可以 improve robustness and achieve fast convergence by using a joint CTC-attention model。