Wind Speed Forecasting via Multi-task Learning

Lencione, Gabriel R.; Zuben, Fernando J. Von

doi:10.1109/ijcnn52387.2021.9534047

Cited by 2 publications

(6 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The number of neurons in the hidden layer, the sampling method of their weights and the adoption of the hyperbolic tangent as activation function are important parameters of the ELM that may have a direct impact on its performance. We follow the same methodology to deal with these aspects as done by [6].…”

Section: B Nonlinear Online Mtl With Extreme Learning Machinesmentioning

confidence: 99%

“…Due to the lack of multi-task time-dependent regression benchmarks, we only test our proposals on the wind speed forecasting dataset proposed by [6]. This case study consists of T = 10 time series of wind speed from wind sites located in Miami, United States.…”

Section: Online Regression Benchmarkmentioning

confidence: 99%

“…This is done as the authors in [6] set, for parsimony purposes, the max length of the auto-regressive vector to be 9 and adopt the AICc criterion to select an optimal length, having this maximum value obtained in most of our preliminary experiments.…”

Section: Online Regression Benchmarkmentioning

confidence: 99%

“…For each subset C k and train/test split percentage µ, we optimize the hyperparameters of the recursive methods on training set and run the online method on test set, as done in [6]. Then, the training set is not used to initialize the parameters of the online methods, which shall be learnt from scratch on test set.…”

Section: Optimization Of Hyperparametersmentioning

confidence: 99%

“…The adoption of nonlinearities in ML methods is intended to offer to models more flexibility to fit data that may not be well represented by linear models. The potential benefits of nonlinear Multi-Task Learning have already been outlined in the literature [6]. This paper comes to fulfill these gaps, with the proposal of two methods for recursive MTL that present better compromise between convergence and computational cost (specially per-instance).…”

Section: Introductionmentioning

confidence: 98%

See 4 more Smart Citations

Online Convex Optimization of a Multi-task Fuzzy Rule-based Evolving System

Lencione¹,

Ayres

Zuben³

2022

2022 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE)

View full text Add to dashboard Cite

This paper introduces two novel approaches for Online Multi-Task Learning (MTL) Regression Problems. We employ a high performance graph-based MTL formulation and develop its recursive versions based on the Weighted Recursive Least Squares (WRLS) and the Online Sparse Least Squares Support Vector Regression (OSLSSVR). Adopting task-stacking transformations, we demonstrate the existence of a single matrix incorporating the relationship of multiple tasks and providing structural information to be embodied by the MT-WRLS method in its initialization procedure and by the MT-OSLSSVR in its multi-task kernel function. Contrasting the existing literature, which is mostly based on Online Gradient Descent (OGD) or cubic inexact approaches, we achieve exact and approximate recursions with quadratic per-instance cost on the dimension of the input space (MT-WRLS) or on the size of the dictionary of instances (MT-OSLSSVR). We compare our online MTL methods to other contenders in a real-world wind speed forecasting case study, evidencing the significant gain in performance of both proposed approaches.

show abstract

Section: B Nonlinear Online Mtl With Extreme Learning Machinesmentioning

confidence: 99%

Section: Online Regression Benchmarkmentioning

confidence: 99%

Section: Online Regression Benchmarkmentioning

confidence: 99%

Section: Optimization Of Hyperparametersmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 98%

See 3 more Smart Citations

Online Convex Optimization of a Multi-task Fuzzy Rule-based Evolving System

Lencione¹,

Ayres

Zuben³

2022

2022 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE)

View full text Add to dashboard Cite

show abstract

Multistep Ahead Multiphase Production Prediction of Fractured Wells Using Bidirectional Gated Recurrent Unit and Multitask Learning

Xiao

et al. 2022

SPE Journal

View full text Add to dashboard Cite

Summary Relying on its strong nonlinear mapping ability, machine learning is found to be efficient and accurate for production prediction of fractured wells compared with conventional analytical methods, numerical simulations, and traditional decline curve analysis. However, its application in forecasting future multistep time series production remains challenging, with complications of error accumulation, growing uncertainty, and degraded accuracy. To this end, we propose a novel multistep ahead production prediction framework based on a bidirectional gated recurrent unit (BiGRU) and multitask learning (MTL) combined neural network (BiGRU-MTL), which can improve prediction performance by sharing task-dependent representations among tasks of multiphase production prediction. The forecasting strategies and evaluation setups for multiple timesteps are elaborated to avoid unfair assessment caused by mixing different prediction confidences over several days. In this framework, BiGRU is in charge of capturing nonlinear patterns of production variation by utilizing both forward and backward sequence information. MTL methods including cross-stitch network (CSN) and weighting losses with homoscedastic uncertainty are incorporated to automatically determine the sharing degree of multiple tasks and the weight ratio of the total loss function. By this means, domain knowledge contained in tasks of multiphase production prediction is deeply leveraged, shared, and coupled to enhance multistep ahead prediction accuracy while meeting the need for multiphase production forecasting. The proposed framework is applied to a synthetic well case, a field well case, and a field multiwell case to progressively prove the feasibility, robustness, and generalization of the BiGRU-MTL model. Experiment results show that the proposed framework outperforms conventional single-task models and commonly used recurrent neural networks (RNNs), furnishing a reliable and stable tool for accurate multistep ahead production prediction. This work promises to provide insights into dynamic production optimization and management in oil- and gasfield sites.

show abstract

Wind Speed Forecasting via Multi-task Learning

Cited by 2 publications

References 28 publications

Online Convex Optimization of a Multi-task Fuzzy Rule-based Evolving System

Online Convex Optimization of a Multi-task Fuzzy Rule-based Evolving System

Multistep Ahead Multiphase Production Prediction of Fractured Wells Using Bidirectional Gated Recurrent Unit and Multitask Learning

Contact Info

Product

Resources

About