Alleviating Sequence Information Loss with Data Overlapping and Prime Batch Sizes

TL;DR

This paper introduces a method to reduce sequence information loss in sequence modeling by overlapping data points and using prime batch sizes, improving performance in text and speech tasks.

Contribution

It proposes a novel approach combining data overlapping and prime batch sizes to mitigate token order imbalance and enhance sequence information retention.

Findings

Achieved state-of-the-art results in text tasks

Improved speech processing performance

Reduced token order imbalance effects

Abstract

In sequence modeling tasks the token order matters, but this information can be partially lost due to the discretization of the sequence into data points. In this paper, we study the imbalance between the way certain token pairs are included in data points and others are not. We denote this a token order imbalance (TOI) and we link the partial sequence information loss to a diminished performance of the system as a whole, both in text and speech processing tasks. We then provide a mechanism to leverage the full token order information -Alleviated TOI- by iteratively overlapping the token composition of data points. For recurrent networks, we use prime numbers for the batch size to avoid redundancies when building batches from overlapped data points. The proposed method achieved state of the art performance in both text and speech related tasks.