Latent Sequence Decompositions

10/10/2016

∙

by   William Chan, et al.

∙

Google

∙

MIT

∙

Carnegie Mellon University

∙

0

∙

share

We present the Latent Sequence Decompositions (LSD) framework. LSD decomposes sequences with variable lengthed output units as a function of both the input sequence and the output sequence. We present a training algorithm which samples valid extensions and an approximate decoding algorithm. We experiment with the Wall Street Journal speech recognition task. Our LSD model achieves 12.9 compared to a character baseline of 14.8 convolutional network on the encoder, we achieve 9.6