Constructing prediction intervals to explore uncertainty based on deep neural networks
Jingling Yang, Liren Chen, Huayou Chen et al. · Journal of Intelligent & Fuzzy Systems · 2024
The conventional approaches to constructing Prediction Intervals (PIs) always follow the principle of ‘high coverage and narrow width’. However, the deviation information has been largely neglected, making the PIs unsatisfactory. For high-risk forecasting tasks, the cost of forecast failure may be prohibitive. To address this, this work introduces a multi-objective loss function that includes Prediction Interval Accumulation Deviation (PIAD) within the Lower Upper Bound Estimation (LUBE) framework. The proposed model can achieve the goal of ‘high coverage, narrow width, and small bias’ in PIs, thus minimizing costs even in cases of prediction failure. A salient feature of the LUBE framework is its ability to discern uncertainty without explicit uncertainty labels, where the data uncertainty and model uncertainty are learned by Deep Neural Networks (DNN) and a model ensemble, respectively. The validity of the proposed method is demonstrated through its application to the prediction of carbon prices in China. Compared with conventional uncertainty quantification methods, the improved interval optimization method can achieve narrower PI widths.
Cited by 0 works. doi:10.3233/jifs-237524
Authors
References
- Scikit-learn: Machine Learning in Python
- Feature selection based on mutual information criteria of max-dependency, max-relevance, and min-redundancy
- Using mutual information for selecting features in supervised neural net learning
- Quantile Regression
- Variable selection using random forests
- Lower Upper Bound Estimation Method for Construction of Neural Network-Based Prediction Intervals
- Construction of Optimal Prediction Intervals for Load Forecasting Problems
- Forecasting carbon price using empirical mode decomposition and evolutionary least squares support vector regression
- Probability density forecasting of wind power using quantile regression neural network and kernel density estimation
- Hybrid Probabilistic Wind Power Forecasting Using Temporally Local Gaussian Process