An empirical study of neural networks for trend detection in time series

TL;DR

This paper empirically evaluates the effectiveness of standard recurrent neural networks in detecting trends in noisy time series, demonstrating their superiority and potential as building blocks for more complex estimators.

Contribution

It provides the first comprehensive empirical analysis of RNNs for trend detection in noisy time series, highlighting their advantages over other methods.

Findings

RNNs outperform other estimators in trend detection

Certain RNN structures show superior versatility

RNNs can serve as foundational components for complex models

Abstract

Detecting structure in noisy time series is a difficult task. One intuitive feature is the notion of trend. From theoretical hints and using simulated time series, we empirically investigate the efficiency of standard recurrent neural networks (RNNs) to detect trends. We show the overall superiority and versatility of certain standard RNNs structures over various other estimators. These RNNs could be used as basic blocks to build more complex time series trend estimators.