Sequence Modeling via Segmentations

TL;DR

This paper introduces a probabilistic sequence model that sums over all possible segmentations, leveraging neural networks for segment modeling, with applications in text segmentation and speech recognition.

Contribution

It proposes a novel segmentation-based probabilistic model with an efficient dynamic programming algorithm, applicable to various sequence tasks.

Findings

Effective in text segmentation and speech recognition

Discovers meaningful segments in sequences

Achieves competitive quantitative results

Abstract

Segmental structure is a common pattern in many types of sequences such as phrases in human languages. In this paper, we present a probabilistic model for sequences via their segmentations. The probability of a segmented sequence is calculated as the product of the probabilities of all its segments, where each segment is modeled using existing tools such as recurrent neural networks. Since the segmentation of a sequence is usually unknown in advance, we sum over all valid segmentations to obtain the final probability for the sequence. An efficient dynamic programming algorithm is developed for forward and backward computations without resorting to any approximation. We demonstrate our approach on text segmentation and speech recognition tasks. In addition to quantitative results, we also show that our approach can discover meaningful segments in their respective application contexts.