Improving the performance of bagging ensembles for data streams through mini-batching

Cassales, Guilherme Weigert; Gomes, Heitor Murilo; Bifet, Albert; Pfahringer, Bernhard; Senger, Hermes

doi:10.1016/j.ins.2021.08.085

Cited by 11 publications

(6 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Since ensemble models with low correlations are preferred in these predictions, the sampling with replacement method allows more difference in the training dataset and, in turn, results in greater differences between the predictions of the base learners. It is worth mentioning that the bagging process, depending on its number of iterations or combination with time series, could be computationally demanding to fit, as explained in [97]. Figure 5 shows a pseudo-code for a bagging NN ensemble algorithm; note that this is a simple example, and the actual implementation of bagging in neural networks may vary depending on each specific case and library.…”

Section: Ensemble Generation Methodsmentioning

confidence: 99%

Ensemble Neural Networks for the Development of Storm Surge Flood Modeling: A Comprehensive Review

Nezhad,

Barooni,

Velioglu Sogut

et al. 2023

JMSE

View full text Add to dashboard Cite

This review paper focuses on the use of ensemble neural networks (ENN) in the development of storm surge flood models. Storm surges are a major concern in coastal regions, and accurate flood modeling is essential for effective disaster management. Neural network (NN) ensembles have shown great potential in improving the accuracy and reliability of such models. This paper presents an overview of the latest research on the application of NNs in storm surge flood modeling and covers the principles and concepts of ENNs, various ensemble architectures, the main challenges associated with NN ensemble algorithms, and their potential benefits in improving flood forecasting accuracy. The main part of this paper pertains to the techniques used to combine a mixed set of predictions from multiple NN models. The combination of these models can lead to improved accuracy, robustness, and generalization performance compared to using a single model. However, generating neural network ensembles also requires careful consideration of the trade-offs between model diversity, model complexity, and computational resources. The ensemble must balance these factors to achieve the best performance. The insights presented in this review paper are particularly relevant for researchers and practitioners working in coastal regions where accurate storm surge flood modeling is critical.

show abstract

Section: Ensemble Generation Methodsmentioning

confidence: 99%

Ensemble Neural Networks for the Development of Storm Surge Flood Modeling: A Comprehensive Review

Nezhad,

Barooni,

Velioglu Sogut

et al. 2023

JMSE

View full text Add to dashboard Cite

show abstract

“…Ensemble modeling can be based on any of the three hypotheses namely bagging, boosting, and stacking. A bagging method lowers the model's variance (Cassales et al, 2021). A boosting method decreases the model's bias (Trizoglou et al, 2021), and stacking enhances the model's predictive power with fine accuracy (Radhakrishnan et al, 2021).…”

Section: Literature Reviewmentioning

confidence: 99%

Real-Time Healthcare Resource Management Planning Using Advanced Machine Learning Methods

Mizan

2024

Preprint

View full text Add to dashboard Cite

The purpose of this study is to develop predictive and optimization models to reduce patients' waiting time for diagnostic tests and medical treatments. Long wait times for receiving medical care is a pressing issue in the Canadian healthcare system. These wait times are spread over different phases of treatment, such as diagnostic tests, physicians' treatments, and appointments with specialists and surgeons. An integrated system to monitor all the phases of treatment along with efficient workload distribution in each phase can reduce patients' waiting time at each of these phases. This study investigates the impact of efficient resource allocation and workload distribution in the medical system. This research also explores the effect of optimized resource allocation on the patients' waiting time in Medicare settings. Resource allocation planning is directly related to the number of patient-arrival, and it is hard to predict such uncertain parameters in the future time frame. The number of patient-arrival also varies across different medical departments and different timeframes which makes the patient-arrival prediction challenging. The goal of this study is to investigate the forecasting effect on patients' waiting time and physicians' workload. To achieve this goal, advanced machine learning technique is integrated with the optimization model. The machine learning technique is used to predict the uncertain parameters of the optimization model for a shorter time span. To predict time-dependent uncertain parameters, such as patient arrival is a major issue, as the prediction may suffer from the concept drift problem. Besides, real-time data are commonly prone to errors due to irregular fluctuations, seasonal biases, and missing values in the data. On the other hand, predicting for shorter intervals requires lower execution time combined with higher accuracy. The developed predictive ensemble model in this research has addressed these issues legitimately with four research contributions. In the first contribution (Chapter 2), we have investigated methodologies for predicting Radiologists’ workload in a short time interval by adopting a machine learning technique. An ensemble model is proposed with the fixed batch training method in this part. To excel in the execution time, a fixed batch training method is used. Secondly, in Chapter 3, an Adaptive Batched-Ranked Ensemble (ABRE) model that reduces the effect of fluctuation using the time-variant windowing technique. Besides, a data aggregation technique is developed and integrated with the offline training phase of the proposed model to tackle the concept drift problem. In the third contribution (Chapter 4), a novel Ensemble of Pruned Regressor Chain (EPRC) method is developed and trained offline to predict uncertain parameters, such as patients’ arrival. Finally, the fourth contribution in Chapter 4, the EPRC method is integrated with a novel multi-objective optimization model to reduce patients’ waiting time, and to determine workload allocation for future timespan. This research enables enhanced decision-making with effective resource allocation and workload scheduling, as well as assists in reducing healthcare expenditure.

show abstract

“…Algorithm 1 process minibatch (...) // As proposed in [Cassales et al 2021] 1: Input: mini-batch B 2: for each trainer T i in trainers T do in parallel ▷ The classification loop 3:…”

Section: Improvement Mini-batching With Loop Fusion For Energy Savingmentioning

confidence: 99%

Improving Performance and Energy Efficiency of the Classification of Data Streams on Edge Computing

Luna,

Cassales,

Senger

2023

Anais Da XIV Escola Regional De Alto Desempenho De São Paulo (ERAD-SP 2023)

View full text Add to dashboard Cite

In this work, we propose a loop transformation to improve performance and energy efficiency of ensembles. We compare the performance of our technique with three other strategies for improving energy efficiency and throughput in data stream classification using six state-of-the-art ensemble algorithms and four benchmark datasets. Our results show that software strategies can significantly reduce energy consumption. Mini-batching improved energy efficiency by 96% on average and 169% in the best case. Likewise, mini-batching with loop fusion improved energy efficiency by 136% on average and 456% in the best case.

show abstract

Improving the performance of bagging ensembles for data streams through mini-batching

Cited by 11 publications

References 20 publications

Ensemble Neural Networks for the Development of Storm Surge Flood Modeling: A Comprehensive Review

Ensemble Neural Networks for the Development of Storm Surge Flood Modeling: A Comprehensive Review

Real-Time Healthcare Resource Management Planning Using Advanced Machine Learning Methods

Improving Performance and Energy Efficiency of the Classification of Data Streams on Edge Computing

Contact Info

Product

Resources

About