ABSTRACT
This study focuses on improving the interpretability of the Long-Short-Term-Memory-Attention hybrid model applied to electrical load forecasting by optimizing its architecture. First, a temporal attention mechanism is added to the Long Short-Term Memory model to understand the temporal patterns learned by the model. Then, we introduce a novel metric assessing the model's interpretability in forecasting, gauging the temporal attention weights' ability to elucidate trends and seasonality. The optimal model architecture is then sought to maximize both interpretability and prediction accuracy, resulting in a Pareto-optimal solution representing the interpretability-accuracy trade-off. Additionally, we investigate the relationship between model architecture and interpretability.
- Maximilian Balandat, Brian Karrer, Daniel R. Jiang, Samuel Daulton, Benjamin Letham, Andrew Gordon Wilson, and Eytan Bakshy. 2020. BOTORCH: A Framework for Efficient Monte-Carlo Bayesian Optimization. In Proceedings of the 34th International Conference on Neural Information Processing Systems (Vancouver, BC, Canada) (NIPS'20). Curran Associates Inc., Article 1807, 15 pages.Google Scholar
- Zachariah Carmichael, Tim Moon, and Sam Adé Jacobs. 2021. Learning Interpretable Models through Multi-Objective Neural Architecture Search. ArXiv abs/2112.08645 (2021).Google Scholar
- Bryce Goodman and Seth Flaxman. 2017. European Union Regulations on Algorithmic Decision-Making and a "Right to Explanation". AI Magazine 38, 3 (Oct. 2017), 50--57. https://doi.org/10.1609/aimag.v38i3.2741Google ScholarDigital Library
- Rui Jiang, Yijia Xue, and Dongmian Zou. 2023. Interpretability-Aware Industrial Anomaly Detection using Autoencoders. IEEE Access 11 (2023), 60490--60500. https://doi.org/10.1109/ACCESS.2023.3286548Google ScholarCross Ref
- Yuening Li, Zhengzhang Chen, Daochen Zha, Mengnan Du, Jingchao Ni, Denghui Zhang, Haifeng Chen, and Xia Hu. 2022. Towards Learning Disentangled Representations for Time Series. In Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (Washington DC, USA), Vol. 1. 3270 - 3278. https://doi.org/10.1145/3534678.3539140Google ScholarDigital Library
- Bryan Lim, Sercan O. Arik, Nicolas Loeff, and Tomas Pfister. 2021. Temporal Fusion Transformers for Interpretable Multi-Horizon Time Series Forecasting. International Journal of Forecasting 37, 4 (2021), 1748--1764. https://doi.org/10.1016/j.ijforecast.2021.03.012Google ScholarCross Ref
- Lkhagvadorj Munkhdalai, Tsendsuren Munkhdalai, Van-Huy Pham, Meijing Li, Keun Ho Ryu, and Nipon Theera-Umpon. 2022. Recurrent Neural Network-Augmented Locally Adaptive Interpretable Regression for Multivariate Time-Series Forecasting. IEEE Access 10 (2022), 11871--11885. https://doi.org/10.1109/ACCESS.2022.3145951Google ScholarCross Ref
- Neeraj, Jimson Mathew, and Ranjan Kumar Behera. 2022. EMD-Att-LSTM: A Data-driven Strategy Combined with Deep Learning for Short-term Load Forecasting. Journal of Modern Power Systems and Clean Energy 10, 5 (2022), 1229--1240. https://doi.org/10.35833/MPCE.2020.000626Google ScholarCross Ref
- Boris N. Oreshkin, Dmitri Carpov, Nicolas Chapados, and Yoshua Bengio. 2020. N-BEATS: Neural Basis Expansion Analysis for Interpretable Time Series Forecasting. In 8th International Conference on Learning Representations, ICLR 2020, Addis Ababa, Ethiopia, April 26-30, 2020. https://openreview.net/forum?id=r1ecqn4YwBGoogle Scholar
- Andreas Theissler, Francesco Spinnato, Udo Schlegel, and Riccardo Guidotti. 2022. Explainable AI for Time Series Classification: A Review, Taxonomy and Research Directions. IEEE Access 10 (2022), 100700--100724. https://doi.org/10.1109/ACCESS.2022.3207765Google ScholarCross Ref
- Ting Wang, Chun Sing Lai, Wing W.Y. Ng, Keda Pan, Mingyang Zhang, Alfredo Vaccaro, and Loi Lei Lai. 2021. Deep Autoencoder with Localized Stochastic Sensitivity for Short-term Load Forecasting. International Journal of Electrical Power & Energy Systems 130 (2021), 106954. https://doi.org/10.1016/j.ijepes.2021.106954Google ScholarCross Ref
- Xiaozhe Wang, Kate Smith-Miles, and Rob Hyndman. 2006. Characteristic-Based Clustering for Time Series Data. Data Min. Knowl. Discov. 13 (09 2006), 335--364. https://doi.org/10.1007/s10618-005-0039-xGoogle ScholarDigital Library
- Yi Wang, Chien-Fei Chen, Peng-Yong Kong, Husheng Li, and Qingsong Wen. 2023. A Cyber-Physical-Social Perspective on Future Smart Distribution Systems. Proc. IEEE 111, 7 (2023), 694--724. https://doi.org/10.1109/JPROC.2022.3192535Google ScholarCross Ref
Index Terms
- Enhancing Interpretability of Electrical Load Forecasting with Architecture Optimization
Recommendations
A comparison of multivariate and univariate time series approaches to modelling and forecasting emergency department demand in Western Australia
The model identification process for VARMA, ARMA and Winters method.Display Omitted VARMA, ARMA and Winters methods are used extensively for planning and management.Multivariate VARMA model is a reliable tool for predicting ED demand by category.It ...
Deep long short-term memory based model for agricultural price forecasting
AbstractAgricultural price forecasting is one of the research hotspots in time series forecasting due to its unique characteristics. In this paper, we developed a deep long short-term memory (DLSTM) based model for the accurate forecasting of a ...
Sales forecasting for Chemical Products by Using SARIMA Model
ICBDE '22: Proceedings of the 5th International Conference on Big Data and EducationSales forecasting is widely used in enterprise resource management, which provides valuable information for efficient management. Sales forecasting facilitates the company to produce and stock products on demand. Based on the analysis of time series, ...
Comments