Comparison of Uncertainty Quantification with Deep Learning in Time Series Regression
This work addresses uncertainty quantification for meteorologists and hedge funds using neural networks in time series regression, but it is incremental as it compares existing methods without introducing new ones.
The paper compared different uncertainty estimation methods for forecasting meteorological time series data to evaluate expectations about uncertainty behavior, such as wider prediction horizons leading to more uncertainty, and found that each method performed variably, partially assessing the robustness of predicted uncertainty.
Increasingly high-stakes decisions are made using neural networks in order to make predictions. Specifically, meteorologists and hedge funds apply these techniques to time series data. When it comes to prediction, there are certain limitations for machine learning models (such as lack of expressiveness, vulnerability of domain shifts and overconfidence) which can be solved using uncertainty estimation. There is a set of expectations regarding how uncertainty should ``behave". For instance, a wider prediction horizon should lead to more uncertainty or the model's confidence should be proportional to its accuracy. In this paper, different uncertainty estimation methods are compared to forecast meteorological time series data and evaluate these expectations. The results show how each uncertainty estimation method performs on the forecasting task, which partially evaluates the robustness of predicted uncertainty.