Translation Between Waves, wave2wave

TL;DR

This paper introduces a novel neural translation model for continuous sensor signal waves, employing window-based representations and iterative back-translation to improve performance on real-world earthquake and activity data.

Contribution

It presents a new seq2seq variant with adaptive wave representation and iterative back-translation for better modeling of continuous sensor signals.

Findings

46% test loss reduction on 1D data

1625% perplexity improvement on high-dimensional data

Effective translation of real-world sensor signals

Abstract

The understanding of sensor data has been greatly improved by advanced deep learning methods with big data. However, available sensor data in the real world are still limited, which is called the opportunistic sensor problem. This paper proposes a new variant of neural machine translation seq2seq to deal with continuous signal waves by introducing the window-based (inverse-) representation to adaptively represent partial shapes of waves and the iterative back-translation model for high-dimensional data. Experimental results are shown for two real-life data: earthquake and activity translation. The performance improvements of one-dimensional data was about 46% in test loss and that of high-dimensional data was about 1625% in perplexity with regard to the original seq2seq.