A Neural Multi-sequence Alignment TeCHnique (NeuMATCH)
Abstract
The alignment of heterogeneous sequential data (video to text) is an important and challenging problem. Standard techniques for this task, including Dynamic Time Warping (DTW) and Conditional Random Fields (CRFs), suffer from inherent drawbacks. Mainly, the Markov assumption implies that, given the immediate past, future alignment decisions are independent of further history. The separation between similarity computation and alignment decision also prevents end-to-end training. In this paper, we propose an end-to-end neural architecture where alignment actions are implemented as moving data between stacks of Long Short-term Memory (LSTM) blocks. This flexible architecture supports a large variety of alignment tasks, including one-to-one, one-to-many, skipping unmatched elements, and (with extensions) non-monotonic alignment. Extensive experiments on semi-synthetic and real datasets show that our algorithm outperforms state-of-the-art baselines.
Cite
@article{arxiv.1803.00057,
title = {A Neural Multi-sequence Alignment TeCHnique (NeuMATCH)},
author = {Pelin Dogan and Boyang Li and Leonid Sigal and Markus Gross},
journal= {arXiv preprint arXiv:1803.00057},
year = {2018}
}
Comments
Accepted at CVPR 2018 (Spotlight). arXiv file includes the paper and the supplemental material